# The AI Survival Guide, by Ace Eddleman A free online book by Ace Eddleman on surviving AI and automation: which jobs are at risk, why this wave is different from past ones, and what to do so you end up on the winning side. Canonical site: https://www.aisurvival.org/ --- # Welcome Source: https://www.aisurvival.org/welcome/ Section: Introduction *[Figure: The AI Survival Guide by Ace Eddleman]* If you're concerned about the future of technology and want to make sure you don't end up a victim of automation, you've come to the right place. AI holds incredible potential, in both good and bad ways. This is a guide for everyone who wants to end up on the winning side. ## Why I Wrote This For a long time, AI was considered a novel toy that nobody—except for the tech-inclined curious—took very seriously. But once it started to make a big splash at places like Google and Facebook, I knew the world would eventually recognize just how big of a deal this set of technologies is. Fast forward to the release of ChatGPT, and suddenly everyone is excited and/or concerned about what AI is capable of. I originally wrote this book back in 2017, but at the time the threat of AI taking jobs seemed a lot more abstract to the majority of people. Now we're all staring dramatic technological upheaval right in the face. What can we do in the face of godlike machines? I do think there's hope, but, as you'll see...we might all be screwed. If we aren't, then this book will be a great resource for you. If we are, well, you can still learn and benefit from what I've put together. --- # Why You Should Read This Source: https://www.aisurvival.org/why-you-should-read-this/ Section: Introduction *[Figure: The AI doesn't want you to read this - just look at that facial expression...]* The reason you should read this book is simple: AI is changing _everything_, but most people are scratching their heads when it comes to figuring out how they can benefit from it. Historically, AI was an incredibly inaccessible field. The only way to get real, hands-on experience with artificial intelligence tech that mattered was to have a PhD and work at a place like Google or Facebook. With the advent of free, powerful tools like ChatGPT, that's now changed. What hasn't changed is the fact that there will be both winners and losers created by this new era of technology, just like every previous epoch in history. The question then becomes: if you don't make any changes, which will you be? ### Will You Get Automated Out of Business? The truth is that the elites with money and influence will be fine, as always. It's everyone else, especially those who don't have the time or resources to keep up with technological changes, who will be affected by the AI revolution the most. With all that in mind, I put this guide together for those of you who don't have deep levels of expertise in this field. You might find out that you're in serious danger of falling behind competitors who know how to leverage AI better than you. Alternatively, you might discover that you can sleep well because it looks like you're good for the foreseeable future. Either way, I guarantee that you'll learn something new about how AI (and automation in general) is going to change your world—and your ability to make a living. ### Avoiding Information Overload To be clear: the problem is not a lack of information. The problem is structure: the information supply is overwhelming, and there aren't enough resources geared towards simplification. My goal here is to remedy that by giving you a roadmap for understanding the problem and, ultimately, solving it in your own life. What you get in this book is not just a tsunami of information to drown in, but a focused set of ideas for understanding how you can avoid losing out in society because of this technological wave. By doing this, I'm giving you some signals to guide your search through our information-rich world. In other words: **It's a way to save you time and energy sorting through the copious amounts of riff-raff that's out there.** ### **Overall Structure** The book itself is a way to build a strong conceptual foundation, rather than a set of answers or predictions. Nobody knows what the future holds, especially when it comes to a world-changing technology like AI. So although I have my own views, I try my best to simply give you the lay of the land in the most accurate way possible and then allow you to leverage that in your own, unique way. Within each section there are a nearly infinite number of rabbit holes you can fall into. How deep you go within each one depends on your time, interest and current level of knowledge. No matter what, you will not find any dead ends here: it's a fast-moving, complex subject that is on a trajectory of exponential expansion. It's also not a detailed explanation of every technical AI concept. There are already so many books and other resources that cover this math-heavy, esoteric field. My goal is to help people who aren't engineers or scientists, not train you to be one (although you will benefit from reading it even if you are interested in that path). This means I've simplified some concepts for the sake of clarity and ignored others because they aren't germane to the subject at hand. ### Leveling the Playing Field One of the biggest problems facing our society is the lack of educational opportunities for people who aren't in their early twenties and living in a tech metropolis. There are boatloads of programs to help people without resources when they're college age, but virtually none for 35 year olds with kids and a mortgage who are facing the threat of job or revenue loss due to AI. Instead, they're expected to just sort of figure it out by Googling around haphazardly. This book is one step towards solving that problem. Don't expect to grasp it all at once. Instead, you should treat this book like it's a class on how to keep yourself alive and thriving in the age of AI. Use it to come up with ideas, execute on at least a few of them, and then come back regularly to find the next piece of the puzzle you want to wrap your mind around. ### Key Points * AI is changing everything at a dizzying pace, and you need to keep up if you want to survive. * This new paradigm will create winners and losers - ask yourself which you'll be and act accordingly. * This book is written to help you avoid information overload within a very deep field. * Read as much or as little as you need to each time you visit, just make sure you apply what you learn so you don't get left behind. --- # We All Might Be Screwed Source: https://www.aisurvival.org/we-all-might-be-screwed/ Section: Introduction Before going any further, it's important to note that we're going to be running on the assumption that [artificial general intelligence](https://en.wikipedia.org/wiki/Artificial\_general\_intelligence) (AI that possesses human-level intelligence) is not going to show up tomorrow, nor will [artificial superintelligence](https://en.wikipedia.org/wiki/Superintelligence) (AI that possesses god-like intelligence far beyond humans). That said, the timeline has shortened dramatically. When I first wrote this book in 2017, most serious researchers put AGI decades away, if they believed it was possible at all. Now, many of those same researchers are talking about years rather than decades. There are two reasons I still focus on sub-AGI automation: 1. Even the most optimistic AGI timelines remain uncertain. Nobody actually knows when or if it will arrive. 2. If and when AGI/ASI is created, it's not unreasonable to assume that every human job (and potentially every human life) is going to be in danger, making this book irrelevant anyway. In other words, AGI remains speculative enough that planning for it specifically isn't useful, but the possibility is no longer something you can dismiss. If you want to know more, I recommend starting with the [Wikipedia article for the "intelligence explosion" concept](https://en.wikipedia.org/wiki/Intelligence\_explosion). The core idea is that an AGI would be intelligent enough to improve itself, and the speed advantages conferred by computer hardware would make that process so rapid that a human-level machine would quickly become far superior to us. In other words, AGI leads to ASI, and it may happen so fast we can't stop it. If AGI can be created and it does lead to an intelligence explosion, we're all screwed—at least in terms of work. Why would anyone ever hire people to produce goods and services again when the machines can do it all? Would the machines even be willing to take on the tasks we give them? There aren't any clear answers to these questions yet. The future of AI is a hotly debated topic, and I'd prefer not to get bogged down in predicting when or if such a powerful technology might show up. With all this in mind, I've decided to focus on the AI that's already here: machine learning, generative AI, and the automation systems being deployed right now. These are already reshaping how we work and live, and we can make better decisions if we focus on technology we can actually observe rather than speculation about god-like machines. But you don't need AGI to be in serious trouble. AI systems available _today_ are already powerful enough to displace millions of workers. That's the threat this book addresses, and as you'll see, it's more than enough to worry about. ### Key Points * Artificial general intelligence (AGI) is human-level machine intelligence. * Artificial superintelligence (ASI) is machine intelligence that is so far beyond ours that it is god-like and unfathomable to our puny human brains. * AGI timelines have shortened dramatically. Serious researchers now talk in years rather than decades, but it remains uncertain. * If human-level AI does show up, there won't be much you or I can do about it since it will likely be too powerful for anyone to resist. * This book focuses on the AI that's already here, which is more than powerful enough to threaten your livelihood without AGI ever arriving. --- # What Is Artificial Intelligence? Source: https://www.aisurvival.org/what-is-artificial-intelligence/ Section: Foundational AI Knowledge *[Figure: AI is narrow. It does one thing extremely well and falls apart the moment the task changes. You are general: you handle problems you have never seen before. That gap is the whole game, and it is closing slowly.]* ## Human Intelligence Humans possess what is sometimes termed **general intelligence**[^1], which is an academic way to say that we're good at dealing with things we haven't encountered before. We know how to _generalize_ what we've learned in the past and transfer that knowledge to new contexts. This is not to say that we're perfect at this process—we aren't—but overall we're incredibly adaptable. We don't fall apart the moment something unfamiliar comes into our world, even if it's alien to us. Humans are capable of learning processes that involve inquiry, experimentation and stored memories to make the most of the novel stimuli we encounter. Our "ecological niche" is pretty much the whole universe, as we are the only animal that is capable of living even in the vacuum of space because of our engineering capabilities. For example, if you visit a city you've never been to before and walk into a restaurant, you should feel confident that you still know how to buy food. Or if you meet a person you've never met before, you may know nothing about them but can generalize based on your past experiences with other people and find a way to get along with them. This comes with trade-offs. Our brains have limits, and those limits create biases, which in turn short-circuit our ability to survive sometimes. But overall, we're quite flexible compared to most other animals. ## Artificial Intelligence So where do machines stand? **Artificial intelligence** (AI) gets thrown around constantly, and it's worth understanding what the term actually means and where it falls short. At the core of modern AI are **machine learning** (ML) systems, which take in data, spot patterns, and then generate outputs that accomplish some task with those patterns. For example, there are ML systems that can identify cancer in x-rays with better accuracy than a person. This is accomplished by feeding an ML algorithm a large set of labelled examples, with some classified as "cancer" and others classified as "not cancer." The ML system then "learns" what patterns constitute cancer and starts to classify real x-rays. Then there are the game-playing systems built by companies like [DeepMind](https://deepmind.com), which utilize a paradigm known as _reinforcement learning_ (RL). An _agent_ (a player) is situated in an _environment_ (a game world, board, etc.) and given a _reward function_ (winning the game, maximizing points, etc.). The agent then plays the game millions or billions of times until it arrives at the best possible strategies. DeepMind's AlphaGo system famously beat the #1 Go player in the world using this approach. And then came the large language models (LLMs): systems like ChatGPT, Claude, and Gemini. These changed the game entirely. LLMs can write code, draft legal briefs, explain quantum physics, and carry on nuanced conversations across virtually any domain. They handle analogies, metaphors, and even irony with surprising competence. Genuinely powerful, genuinely useful, and a qualitative leap beyond older ML systems. **But they are still not human intelligence.** LLMs don't experience the world. They don't have goals of their own, and they work in fundamentally different ways than our brains do. An LLM generates its outputs through sophisticated pattern completion over enormous datasets. Think of it as an alien form of capability: it can look like understanding without working the way understanding works in a human brain. This distinction matters less than it used to. Whether or not these systems are "truly intelligent" in a philosophical sense, they are capable enough to reshape entire industries. What matters for survival is a more practical question: what can it do, and what can't it do? If you can answer that accurately, you're ahead of most people. ## Why Knowing This Matters There's tons of hype around AI these days, and it's easy to get "lost in the sauce" of that hype. But it's equally dangerous to dismiss these systems as toys. The truth is somewhere in between, and finding it requires understanding what's actually happening under the hood. Modern AI systems like ChatGPT are built on top of **large language models** (LLMs), which ingest oceans of data and learn to generate outputs based on patterns in that data. The results are often impressive, sometimes astonishingly so. These systems can write working software, pass bar exams, solve graduate-level math problems, and generate analysis that would take a human researcher hours. But they also have real limitations. They can confidently produce nonsense (a phenomenon called "hallucination"). Sometimes they fail at tasks that seem trivially easy to a human. They lack lived experience or genuine understanding of consequences. And they can be manipulated in ways that a thoughtful person wouldn't be. Overestimate AI and you'll trust it with things it can't handle, making costly mistakes. Underestimate it and you'll be blindsided when it takes over tasks you thought were safe. People who thrive will be those who develop an accurate, clear-eyed view of what these systems can and can't do. That picture is changing fast, which is why staying informed matters so much. ## A Note on Terminology Throughout this book I use "AI" and "ML" somewhat interchangeably. Technically, machine learning is the set of techniques that powers modern AI, but in everyday conversation people just say "AI." So will I. When a more specific distinction matters, I'll call it out. ## Key Points * Human intelligence, sometimes called general intelligence, is about the ability to understand the world, generalize knowledge and solve novel problems. * Modern AI is genuinely powerful. It can write code, pass professional exams, and handle tasks across many domains. * But it is not human intelligence. It works in fundamentally different ways and has real, significant limitations. * What matters for survival is a practical question: what can it do and what can't it do? That picture is shifting fast. * Both overestimating and underestimating AI are dangerous. Develop a clear-eyed understanding of its actual capabilities. [^1]: _It's worth mentioning that the concept of general intelligence is controversial, particularly when considered in the context of other animals. An octopus may not be as adaptable or "generally intelligent as us," for example, but they are masters of their environments. Does that mean they have low general intelligence, or is general intelligence a poorly-defined idea that is too human-centric? This is an ongoing debate._ --- # A Brief History of Automation Source: https://www.aisurvival.org/a-brief-history-of-automation/ Section: Foundational AI Knowledge *[Figure: Every wave of automation was going to be the big one. Looms, steam, the assembly line, computers, the internet. Each reshaped work and moved on. Read it as a timeline: the recent waves are bunching up, arriving decades apart now instead of centuries.]* As a general principle of capitalist enterprises, every business is searching for a way to remove people from their operations. People are expensive, temperamental and difficult to control. They ask for vacations, slack off, sexually harass each other and generally cause chaos for companies. Automation gives businesses what they want: lower costs, higher profits, increased productivity and far fewer headaches. Employers have understood this dynamic since the first machines arrived during the Industrial Revolution, and they've been utilizing automation ever since. This has caused tensions with the people who are the targets of layoffs, and it also creates wider economic problems because it's people who buy things. If those people don't have incomes, they can't buy as much, which means those same companies can't make as much money. ### The Historical Pattern of Automation and Adaptation What's a business, or a worker, to do? Historically, the overall (highly simplified) pattern looks something like this: * New invention radically alters the balance of power between workers and owners. * Owners rapidly replace or downgrade workers with the leverage this new invention gives them. * Workers resist in the near-term (example: [the Luddites](https://en.wikipedia.org/wiki/Luddite)), sometimes resulting in violence. * Businesses start to recognize limitations in the technology or realize they have to soften their stance to keep the money flowing. * Workers pick up skills that work for this new arrangement and start to replace their lost incomes. * Both sides adapt, and a new equilibrium is created...until the next invention shows up. ### A Better Long-Term Solution In previous eras, automation meant that jobs were replaced, but (in the long term, at least) it worked out for everyone. More jobs were created than replaced in the long run, and automation has undoubtedly saved countless hours of human labor from being wasted on pointless, repetitive tasks. Examples of this type of automation are all over the place. During World War 2, people (mostly women) [manually calculated](https://en.wikipedia.org/wiki/Human\_computer) shell trajectories, ideal bombing altitudes and all kinds of other data points needed for battle. That was a boring, difficult job and the inevitable errors that come from having people doing that kind of work by hand put lives at risk. If you wanted to get cash out of a bank, it used to be a matter of dealing with a bank clerk, who had to do a whole bunch of manual writing and other...clerky...tasks before they could hand you your money. Then the ATM was created, which replaced that entire cycle of tasks for the people who worked at the bank. Same goes for people who pumped gas, used pickaxes in mines, and on and on and on. Automation has reshaped labor as technology progresses, with predictable responses from those who benefit and those who are hurt by it. ### Not So Great in the Short-Term The ugly truth is that the Luddites were largely correct about their short-term prospects: the textile machines designed to replace them did, in fact, replace them. There weren't any social safety nets, and anyone who had dedicated a large block of their lives to this work was in serious trouble. Their response (smashing machinery and creating chaos for factory owners) was a rational one. They faced an existential crisis and responded in the only way that seemed effective. Unfortunately, they also lived in a time where the (British) government had zero issues with killing protesting workers on the spot, then arresting any survivors. All this being said, it's for the best that textiles were automated. It allows for much faster, cheaper production and using people to do something that is so easily done by a machine is a waste of human potential. Eventually, textile workers found other ways to make a living and the world moved on from the Luddite cause. This is the great balancing act we've had to play throughout history: workers get hurt (sometimes quite badly) in the short-term by increased automation, but long-term it tends to be for the best. ### Humans Should Not Do Boring Jobs Automating that kind of work is **a good thing**, even if it puts people out of a job. Can anyone seriously argue that their life would be enhanced if they had to pump gas all day? No, of course not. But it kept food in some people's fridges, and that's why they were viewed as important at all. Rather than spending our time raising crops and assembling widgets by hand, we can now engage in work that means something to us. I think we can all agree that this has been a positive development. A larger problem is not technological, but political. When governments and businesses feel no obligation to take care of workers who have been made redundant by new technologies, it creates a wide variety of serious, society-wide conflicts. This is a much larger discussion that is beyond the scope of this guide, but the fact remains that technology is never neutral: it creates winners and losers. Who falls into each category and how the victors treat the vanquished both help determine whether a technological transition is a net positive or a net negative. ### Enter Computers Fast forward a bit from the days of the Luddites, and computers have continued this same dynamic. When we started to utilize computers for tasks that only require calculation (not creativity), we made our collective lives better because we could use our minds for things besides simple number crunching. This style of automation—taking people out of the picture for tasks that require high speed and precision—has been the norm since the dawn of the Information Age. If the primary constraint on people is speed and/or precision, computers have largely taken their place. The only limitation on automating like this is access to computing power, and that problem has been solved. Once you have the CPUs to do the task, it's just a matter of defining a clear set of steps (aka an [algorithm](https://en.wikipedia.org/wiki/Algorithm)) and putting it all into place. ### Familiar Math Knowing all this, it makes sense that people have been taken out of factories and factory-like environments. A very simple robot can do the same thing over and over again for a fixed cost and at high levels of precision and speed. Robots in a factory can be given a small number of well-defined tasks and do those tasks much faster and better than a human. Even better (from a manager's perspective, at least), is the robots don't demand raises, ask for breaks or go on strike if they don't get paid enough. Fortunately, computers were such a powerful general-purpose technology that they opened far more doors than they closed. A worker fired from a manufacturing plant can go online to search for other jobs, learn new skills, network, and so on. Losing your job to that kind of automation isn't the end of the world. ### An Ominous Future Anything that involves crunching numbers at crazy high speeds is the low-hanging fruit of computing. And that low-hanging fruit has been eaten for the most part. There will always be new applications for this kind of technology popping up, but as time goes on the returns get smaller and the capital required to take advantage of opportunities balloons. It used to be that a person or two in a garage could use a computer to create a world-changing technology company. While there's always the off-chance something like that could still happen, this dynamic was possible because the aforementioned low-hanging fruit was still on the branch. Now taking advantage of tech-related opportunities, at least ones that involve actual innovation, requires more time, more money and more people. As with most things in human history, this sort of equilibrium could not last forever. Disruptive technology inevitably shows up to change the landscape, and that's exactly what happened with the arrival of machine learning (ML). _Please note: This is a HIGHLY simplified version of events. There is a library worth of books out there about this and it's a supremely political subject, so don't crucify me for not including more details._ ### Key Points * There is a historical tension between business owners and workers over automation. * Business owners want to get rid of workers to pump up their bottom lines. * Workers don't want to get fired because they don't generally like to starve or be homeless. * Examples like those of the Luddites in Britain illustrate that workers have good reason to be wary in the short-term, but overall society is better off in the long-term. * Computers continued this dynamic but carried it over to tasks that required far more speed and precision than previous eras. * As with previous eras, computers have given workers new ways to make a living even if they get replaced by machines. * The low-hanging fruit of the Information Age is gone. * What's the next stage in the evolution of this story? Machine learning. --- # The Rise of Machine Learning Source: https://www.aisurvival.org/the-rise-of-machine-learning/ Section: Foundational AI Knowledge The first wave of Information Age automation involved programmers manually defining everything their computer systems should do. As discussed in the last chapter, this fits well into the model for all previous eras of automation, where a human designs a piece of technology to make a machine that does things faster, cheaper, etc. Machine learning (ML) turned all of that on its head. While the old way of automation focused on throwing processing power at simple, well-defined tasks, machine learning takes a different approach. The old way's big limitation in the age of cheap, abundant processing power is the ability to define and execute a set of steps (an algorithm). If the task is straightforward, like assembling a car bumper, then that's not particularly hard to get that algorithm up and running. That approach runs into problems in domains where there's a large amount of variability and a substantial number of choices that have to be made on a regular basis. This presented problems in the past because old methods relied on the limited brains of programmers, who can only account for a small number of possibilities for any given activity. Constructing logical constructs of how to do something becomes untenable quickly for all but the most basic activities. Enter [**machine learning**](https://en.wikipedia.org/wiki/Machine\_learning) (ML). Rather than depending on programmers determining all possible paths ahead of time, ML uses data to generate programs that solve problems on their own. You can think of ML like a big, black box that takes in data (as much as you can give it) and spits out programs. There are a bunch of different ways to generate these programs, but the type of ML creating the most economic value right now is a set of techniques known as [**supervised learning**](https://en.wikipedia.org/wiki/Supervised\_learning). Supervised learning revolves around using data to find patterns, and then building programs that utilize those patterns in some way. Another popular set of techniques is [**reinforcement learning**](https://en.wikipedia.org/wiki/Reinforcement\_learning) **(RL)**, in which agents maximize some kind of reward while operating within a specific environment. RL has been at the core of the most famous AI game playing systems, such as DeepMind's AlphaGo and AlphaGo Zero. The specifics of those particular topics are very math-heavy and not particularly relevant for right now, so we're not going to zero in on the details. For now, just remember this: **ML is all about using data to infer patterns, create models around those patterns and then generate some kind of automated output.** ### ML Out in the Wild An example that you've probably already used today is the way that Google suggests things when you use the search bar. The suggestions you get were not manually entered by a programmer—that would be far too labor-intensive. Instead, Google utilizes machine learning to constantly scan their incoming search data and spot patterns in the search terms. If you go to Google right now and type in "What is..." you'll see some suggestions, which are likely not related to what you're looking for. That's because that phrase is extremely broad. Once you narrow it down a little bit, like "What is machine" you'll get closer to what you want. Google can do this because their ML algorithms recognize that those combinations of words tend to lead to specific searches. As more and more people use their search engine, those algorithms automatically build those suggestions and give people increasingly relevant results. The only reason Google can do this so effectively is because they deal with enormous amounts of data. On any given day, they're processing [40,000 searches _per second_](http://www.internetlivestats.com/google-search-statistics/). If Google was just starting out and processing only a few searches a day, then they'd have to wait a long time before they had enough data to use ML. That's because ML systems utilize statistics to spot patterns, and it's very difficult to accurately spot patterns with small sample sizes. ### Building Houses vs. Growing Plants ML technology is so powerful because it allows programmers to _teach_ software how to do things, rather than strictly define what software should do. Once data is available, ML algorithms can continuously improve themselves and generate increasingly complex programs. There's an element of mystery to the whole ML process as a result of this, because the algorithms are generating programs on their own. Programmers are no longer specifying all the moving parts of their software, and are instead simply telling their ML systems what kind of end results they want (better searches, improved movie recommendations, etc.). Because they aren't designing the output programs, they often have no idea exactly how they work. As crazy as that may sound to those outside of the tech world, this is the norm when you're dealing with massive amounts of data and throwing them into ML algorithms. It's a big, extremely profitable set of black boxes. Take, for example, self-driving cars. Even though Google and several other companies have built vehicles that can drive themselves, nobody knows exactly how they do it. This is because self-driving cars use ML to _teach_ themselves how to drive, and it's a process of adaptation rather than explicit instruction. First, the engineers working on these projects design a set of target outputs. Their primary goal is to have the car drive itself **safely** (an important distinction, since a car could get to its destination in a variety of very unsafe ways), so the goal is to have a program that knows when to stop at stoplights, make U-turns, merge onto the freeway, and so on. Engineers then slap a plethora of sensors onto their prototype vehicles, and then allow them to drive around. When the vehicle makes a mistake (like stopping when it shouldn't), it looks at the data surrounding that mistake, updates its internal algorithm, then tries again. While the engineering team is not entirely hands-off (they have to regularly make tweaks to the ML systems to ensure the targets are understood and correct), the task of driving is not defined by them. They simply tell the ML systems what outputs they want, and the car figures out the rest. The differences between traditional software systems and ML systems can be compared to the difference between building a house (old way of building software) and growing a plant (new way of building software). A house is built by laborers from blueprints that are created by an architect. The laborers are expected to follow the design laid out in those blueprints exactly, and they fully understand what the end product will be because of the nature of that construction process. Growing a plant, on the other hand, involves placing a seed into soil and then tending to it until it turns into a bush, a tree, a tuber, and so on. How well it turns out depends on how much water you use, what nutrients you put into the soil, etc. You don't have to understand everything about that growth process to get what you want out of the plant, but you do need to provide it with the right environment to thrive. That's ML in a nutshell: you give an ML system some data, and see what comes out. It's a completely different mindset than what most people are used to, but it's becoming increasingly popular as AI gets more advanced. ### The Generative AI Explosion Growing plants becomes an even more apt analogy with the latest generation of ML: **generative AI**. Systems like ChatGPT, Claude, and Midjourney are trained on massive datasets (in some cases, essentially the entire internet) and can generate text, images, code, music, and video that is often indistinguishable from human-created content. These systems represent a qualitative leap from earlier ML. A cancer-detection algorithm could only do one thing. A self-driving car could only drive. But a large language model can write a legal brief, then debug a Python script, then explain quantum physics to a ten-year-old, all in the same conversation. That generality is what makes generative AI so disruptive. It touches nearly every knowledge-work profession at once, rather than picking off one occupation at a time like previous waves of automation did. Implications of this shift are explored in the rest of this book, but the key insight is simple: **ML has gone from automating narrow, well-defined tasks to automating broad, creative, and cognitive work.** That changes the survival calculus for almost everyone. --- # The Role of Data Source: https://www.aisurvival.org/the-role-of-data/ Section: Foundational AI Knowledge *[Figure: The Bitter Lesson. Every time researchers stopped hand-coding their knowledge in and just threw more compute and data at the problem, the machines pulled ahead. It stings our egos, and it keeps being true.]* The examples from the last section highlight a critical point about automation that is worth repeating: it requires boatloads of data. A company like Google can successfully use ML because they have access to massive datasets about what their users are looking for when they use their search engine. If you don't have data, you can't work with AI. Data is behind the success of many other large companies. Here are just a few examples: * Netflix uses a massive trove of user data and ML to determine what movies and TV shows to recommend to its users. * Amazon utilizes its large datasets about buying behavior to figure out how to entice people to add more items to their shopping carts. * Facebook and Instagram deliver highly targeted ads based on each user's individual "liking" behavior. However, there are some surprising truths about how data is used in the AI world that are worth considering. For one, it's no longer a matter of having clean data. OpenAI's GPT systems, although still flawed, have brought high-level AI tech to the masses by simply ingesting everything on the internet. The generational differences between GPT-2 and GPT-3, for example, were purely based on how much data was fed into each - the more data, the better the results, apparently. ### The Bitter Lesson There's a famous essay called[ The Bitter Lesson](http://www.incompleteideas.net/IncIdeas/BitterLesson.html) by Richard Sutton which lays this out. It's controversial in the AI world, but the gist of what he wrote is this: every time we just scale up our computing power and data, we get better results than when we try to incorporate our own knowledge into these systems. It's a "bitter lesson" because we take it as a hit to our ego when a computer just gobbles up a bunch of noisy data and outperforms AI that humans have painstakingly tried to infuse with our collective knowledge. That's been true with a wide variety of systems, from Deep Blue to ChatGPT. Just throw a bunch of processors and a massive amount of data at a problem, and boom, you get an AI system that outperforms anything we build by hand. ### Proprietary Datasets Why am I telling you all of this? Because there's a race on now in the world of AI, to find and leverage specialized datasets. People in the field originally thought that data would be the prime driver of monopolization in AI, but that might not be the case—if all you need is a ton of data, then just hoover up the entire internet! Nobody can own that, and it commodifies anything that leans on that dataset. In some sense, it will create a situation where everyone is using pretty much the same dataset, with maybe a few minor differences. That's good news for consumer-facing AI, because it means it's very unlikely anyone will own it all (hopefully). That leaves only one type of data with value: highly specialized data that can be used for [_fine-tuning_](https://en.wikipedia.org/wiki/Fine-tuning\_\(deep\_learning\)). Anybody who wants to build a product that just depends on the entire internet's data will, for reasons outlined above, run into problems. But a company (or an individual, in some cases) can generate a competitive edge for themselves by training their AI on proprietary datasets. Once they have that data, they can train it through the fine-tuning process to do specific tasks at levels that often far exceed what a human is capable of. Even if it's only average compared to humans, the sheer volume of outputs that they can generate means these algorithms are extremely valuable. The companies I mentioned before will still have quite a competitive advantage, for example. Consider that Google still has access to their own search results, which is a massive proprietary dataset that is worth billions of dollars. Likewise for Netflix, Amazon, Facebook, and so on. As you move deeper into the world of AI, ask yourself: what kind of data do I have access to that nobody else does? Much of the economic opportunity moving forward (at least in industries where this tech can play a role) will be in these datasets. You can either use it for your own purposes (such as pumping out specific types of content) or sell it to somebody else for their ML algorithms to gobble up. ### Summary * Data is the core component of all AI systems. * It appears that sheer volume of data and computational power are enough to beat humans in many AI applications. * The best way to create an edge with AI is to train it on a large, proprietary dataset and then fine-tune it. --- # Are You Replaceable? Source: https://www.aisurvival.org/are-you-replaceable/ Section: Foundational AI Knowledge Given what you know about AI, ML and the importance of data, we need to define a framework for spotting occupations that are likely to get automated. On the one hand, as an entrepreneur, it may mean your barrier to entry is lower. You won't have to hire as many people (or any, in some cases), and that equals more potential opportunity. For employees, the situation is a bit more grim, since your paychecks could end up on the chopping block. And it won't be a matter of finding a new employer: you'll need to come up with a whole different career to pursue! Anyone who is mid-career or later will have to think long and hard about this subject. Here is a simple set of questions that you can use as a heuristic for figuring this out: 1. Does the work involve predictable patterns of activity?\ _If it's very predictable, then it's a prime target for all kinds of automation._ 2. Are there already frameworks in place for handling the work?\ _Frameworks often translate into structures that can be fed into an algorithm._ 3. How difficult is it to gather data about what is being done?\ _If it's easy to gather data, it may be easy to train a machine learning system._ In case you aren't sure about whether data is being (or even can be) collected, think about how many machines and/or pieces of software you interact with on a daily basis. This is important, because jobs that involve lots of tech are jobs where data can be easily collected. A related question is: **"How ambiguous is my work?"** Even if there are lots of data points available, a job can be very difficult to automate when there's a lot of human-induced ambiguity involved. These types of jobs are usually heavy on human-to-human interaction, and tend to revolve around systems that are not easy to define. ### What Can Be Automated Given this framework, let's walk through some examples and see if we can spot some potential targets for automation. #### Trucker: Eventually, But Slower Than Expected Trucking is a multi-billion dollar industry that relies heavily on human labor. Companies need to move physical goods across land, and trucking does that by putting a human being behind the wheel. For years, self-driving trucks were presented as an imminent revolution. And yet, progress has been slower than almost everyone predicted. The technology works reasonably well on highways in good conditions, but the messy reality of urban driving, bad weather, construction zones, and unpredictable human behavior has proven incredibly difficult to solve. This is actually a great illustration of Moravec's paradox (more on that below): driving looks like a simple, mechanical task, but it's actually deeply embodied and requires the kind of real-world judgment that AI still struggles with. The data is easy to collect, the task is well-instrumented, and the financial incentive is enormous. And _still_ it's taking longer than expected. That said, the trajectory is clear. Autonomous trucking on major highways is being deployed in limited corridors, and the technology will expand over time. Truckers should be preparing for a future where their role changes significantly, even if full automation isn't arriving tomorrow. #### Fast Food Worker: Being Automated in Stages Fast food is a ripe target for automation because it's an assembly line for food. Getting food prepared and delivered to the customer is a well-defined, predictable, and repetitive process. Ordering has already been largely automated. Self-service kiosks and AI-powered drive-through systems are now common at major chains. McDonald's, Wendy's, and others have been deploying these for years, and the trend is accelerating. Food preparation has been slower to automate (another instance of Moravec's paradox). While the process _seems_ simple (grab buns, add condiments, drop fries, fill a cup), the physical manipulation involved is actually harder to automate than the cognitive work of taking an order. Robot kitchens exist, but they're expensive and brittle compared to a human who can handle the thousand small variations that come up in a real kitchen. The pattern here is instructive: the _information-processing_ parts of the job (taking orders, managing inventory) are getting automated first, while the _physical_ parts lag behind. This is the opposite of what most people would expect, and it's a pattern you'll see across many industries. #### Software Engineer: Rapidly Being Transformed When I first wrote this book in 2017, I predicted that programming would be automated and caught flak for it. Most programmers were confident their jobs were safe. They were wrong. AI coding tools have gone from novelty to near-necessity in just a few years. What started with GitHub Copilot suggesting lines of code has evolved into AI systems that can build entire applications from natural language descriptions, debug complex codebases, and handle the kind of routine programming work that used to employ armies of junior developers. Why did this happen so fast? Exactly what my original framework predicted: there's an enormous amount of data about how programs are built (open source repositories, documentation, Stack Overflow), and most programming involves assembling well-known components in predictable patterns. This made software engineering one of the most data-rich, pattern-heavy professions in existence. A prime target for ML. Software engineering is now shifting from "person who writes code" to "person who directs AI to write code and verifies the output." The people who thrive will be those who understand systems at a deep architectural level, can evaluate AI-generated code critically, and can handle the genuinely novel problems that AI still struggles with. Polarization has arrived. Elite engineers, particularly those working on AI systems themselves, command enormous salaries. But the market for mid-level developers who primarily write routine code is getting squeezed hard. If your value proposition is "I can build a standard web app," you're competing against tools that can do it in minutes for pennies. #### Copywriter: Already Being Automated People who write the words you see on ads, landing pages, and other forms of marketing materials are quickly seeing their ability to survive in the marketplace disappear. With the appearance of ChatGPT and a variety of marketing-specific tools built on the OpenAI GPT systems, it's become trivial to generate this form of writing. For now, the role of a copywriter has become copyeditor - they feed some ideas into the machine, and then edit or add to the output. While they can certainly still add some value in places, it's hard to argue that becoming a copywriter or hiring a full-time copywriter is a wise decision at this point. #### Lawyer: More Threatened Than Expected I originally put lawyers in the "probably safe" category. That call is looking shakier than I'd like. AI is now doing far more than just sorting through court rulings. LLMs can draft contracts, summarize case law, generate legal briefs, and even provide basic legal analysis that used to require hours of attorney time. The paralegal and junior associate work I said would get automated? It's happening faster than anticipated, and the automation is creeping further up the skill ladder than most lawyers expected. That said, the core of legal work remains deeply human: consulting with clients in sensitive situations, crafting a narrative from incomplete evidence, reading a jury, negotiating with opposing counsel. These tasks are drenched in ambiguity, interpersonal judgment, and the kind of contextual awareness that AI still can't match. The likely outcome is fewer lawyers doing more work, augmented by AI tools. Attorneys who learn to leverage AI effectively will be far more productive, which means firms need fewer of them. If you're in law, one lawyer with AI tools can now do the work of three without them. Guess which two get cut. #### Police Officer: Probably Won't Be Automated Like lawyers, police officers have a job that's incredibly difficult to automate. This is because their work revolves around complex, ambiguous interactions between people. Couple that with an equally complex, ambiguous legal system which has to be followed and you'll quickly see how difficult it would be to build robot cops. Consider how hard it would be to automate the most controversial (and yet central) component of the job: the use of force. Police officers are unique in that they have to make decisions–usually very quickly–about whether it's acceptable to pull out their guns and shoot other human beings. Their decisions are sometimes wrong (for a variety of reasons) and [innocent people die](https://en.wikipedia.org/wiki/Shooting\_of\_Philando\_Castile). Even though there are certainly cases where the officer is acting in bad faith, much of the time these wrongful shootings are due to the need to make split-second decisions about ambiguous situations. Each situation is unique, and it would be incredibly difficult to build a computer system (even an ML system) that could accurately detect all the nuances needed to make a "good" decision to shoot. Officers have been, and will continue to be, augmented with technology (such as cameras on their cars and uniforms), but the core of their job will be difficult to automate for the foreseeable future. ### Moravec's Paradox An interesting idea that has proven to be quite accurate over the years is [Moravec's paradox](https://en.wikipedia.org/wiki/Moravec's\_paradox). The TLDR of this concept is that we're quite bad at building machines which do things we view as easy, but good at building machines that handle tasks we view as difficult. Consider this: unless you're an infant (and if you're reading this as a baby, I'm impressed), walking, talking and grabbing things aren't tasks you have to think too hard about. We do them subconsciously, as they are highly evolved abilities that have been around since the beginning of our species. Nobody has to really teach you how to do any of those things, unless you've suffered some kind of injury and have to re-learn them. On the other hand, playing chess and writing code are _not_ evolutionarily ancient skills. As such, you have to be trained how to do those things and many people find both tasks very difficult. Because they don't come naturally to us, we must be taught how to do both. Which one is harder to create a machine around? Turns out it's the first category, largely because we aren't even sure how to describe those kinds of actions to a computer. We've poured who-knows-how-much time and money into building robots that can walk like humans, but after decades of trying we still haven't nailed it. Translate this to work, and you can see where the pattern is emerging. Blue collar jobs, which everyone thought would be the first to go, tend to be the hardest to automate. How can you create a machine that crawls under a sink and does the work of a plumber when just about every house and plumbing situation has its own weird wrinkles? White-collar jobs like designers and writers are losing their jobs to machines rapidly, despite the fact that everyone thought "creative" jobs would be the safest. Turns out that creativity isn't what matters, it's ambiguity. ### Augmentation and Displacement AI is both augmenting and displacing workers, often simultaneously. The pattern that's emerging is that AI augments the best performers in a field while displacing the rest. One lawyer with AI tools can do the work of three without them, which means two of those three are out of a job even though the work is still being "done by a human." Survival in this phase of the technology requires you to be on the right side of that equation: the person who uses AI tools to multiply their value, not the person who gets multiplied out of a job. ### Summary * You can ask three questions to get a rough idea about how in danger a given profession is: Does the work involve predictable patterns of activity? Are there already frameworks in place for handling the work? How difficult is it to gather data about what is being done? * Jobs with high levels of ambiguity and uncertainty are the least likely to be fully automated, while those that are structure-heavy and prone to data collection are the most likely. * Moravec's paradox describes this dynamic well: physical and embodied skills are proving far harder to automate than cognitive, information-processing tasks. * AI is increasingly a force for both augmentation and displacement. The pattern emerging is that one person with AI tools can do the work of several without them. --- # Learn More About AI Source: https://www.aisurvival.org/learn-more-about-ai/ Section: Foundational AI Knowledge Now that you have a very high-level view of how modern AI works and how it might threaten your economic survival, you can go a bit further into the rabbit hole without getting lost. Understanding the threat you're facing is the first step towards making sure you can survive and thrive in a rapidly changing world. However, I don't want this particular book to get bogged down in such a deep, complex subject. Instead, I'm going to give you some learning materials and I expect you to go through them. You can't respond to AI appropriately if you don't have a baseline understanding of how it works, and these books and videos will help build that foundation. It's also important that you don't just go through these resources and stop there. Staying on top of developments in tech is a key behavior in the fight to stay relevant in an increasingly automated world. It's easier to protect your livelihood if you can spot incoming trends that might threaten your position. All of these recommendations are based on the assumption that you aren't already well-versed in the math and technology behind automation, so don't be intimidated if you aren't technical. ### Read These [**The Master Algorithm**](http://amzn.to/2f5jnJ6) **by Pedro Domingos** This is by far the best explanation of machine learning for people who are not involved in actually building machine learning systems. If you want to understand what can and can't be automated, this is an excellent resource. Written before the LLM era, but the foundations it covers are still how everything works under the hood. [**The Future of Employment**](http://www.oxfordmartin.ox.ac.uk/downloads/academic/The\_Future\_of\_Employment.pdf) **by Carl Benedikt Frey and Michael A. Osborne** This is a fairly long research paper about the potential impact of automation on jobs in the future. Like this book, it bases its assumptions about automation on the idea that we're just going to maximize ML, which still leads to some startling conclusions. [**Co-Intelligence: Living and Working with AI**](https://www.amazon.com/Co-Intelligence-Living-Working-Ethan-Mollick/dp/059371671X) **by Ethan Mollick** Mollick is a Wharton professor who has been one of the most thoughtful voices on how AI will reshape work. This book is practical, grounded, and avoids both hype and doomerism. Essential reading for understanding how to actually work alongside AI systems. [**The Coming Wave**](https://www.amazon.com/Coming-Wave-Technology-Twenty-first-Centurys/dp/0593593952) **by Mustafa Suleyman** Written by a co-founder of DeepMind, this book lays out the case for why AI (alongside other technologies) represents a fundamentally different kind of challenge than what humanity has faced before. A good complement to the more economics-focused resources. ### Watch These [**The Rise of the Machines – Why Automation is Different this Time**](https://www.youtube.com/watch?v=WSKi8HfcxEk) **by Kurzgesagt – In a Nutshell** Even if you don't want to read any books, at the very least watch this video. It covers why this wave of automation is not like the automation of the past, which is what I've been talking about throughout this chapter. [**Humans Need Not Apply**](https://www.youtube.com/watch?v=7Pq-S557XQU) **by CGP Grey** This is a 15-minute exploration of the economics and potential impact of automation. Another must-watch for anyone who's concerned about being automated. Made in 2014 but still highly relevant. If anything, the trends it predicted have accelerated. [**A Friendly Introduction to Machine Learning**](https://www.youtube.com/watch?v=IpGxLWOIZy4) **by Luis Serrano** A gentle thirty minute explanation of the most popular algorithms in machine learning. After watching this, it will be easier to spot which types of ML are being used out in the wild, even in products you use every day. This video is more technical than the others, but you won't regret watching it. --- # Implications of AI Source: https://www.aisurvival.org/implications-of-ai/ Section: What This All Means All that information is well and good, but it's just as important that you grok the implications of what it all means. This is a survival guide, after all, and not a PhD thesis. The short answer: it's not great for the vast majority of those who work for a living. As I mentioned before, AI will create winners and losers—and workers appear to be losing already. Let's find out why... --- # Job Scarcity Source: https://www.aisurvival.org/job-scarcity/ Section: What This All Means *[Figure: Automation shrinks the pool of paid work while more people compete for what is left. Two lines heading in opposite directions, and you are standing in the gap between them.]* There is a high risk that jobs in general will become far more scarce and therefore competitive. Unlike previous eras, where labor demand shifted from one field to another, this wave of automation has the potential to remove a nontrivial number of jobs from the economy—permanently. Because businesses always have an incentive to reduce the number of humans on the payroll, workers will continue to face an uphill battle as AI becomes more incorporated into the workplace. As it stands today, the technology already exists for one person to start and run a successful business, using only cheap tools and contracted labor. That trend appears to be set to accelerate, meaning companies will find more and more ways to avoid hiring people as full-time employees. The ongoing "gigification" of the economy, where more and more people find themselves forced to sell their labor in small, cheap bursts on platforms like Uber and Instacart, seems likely to accelerate. What looks to be on the horizon is what's called _polarization_: a small subset of jobs will become extremely well-paid, and everyone else will be stuck driving Ubers to make ends meet. The jobs that pay well will be those that are either directly related to AI (such as machine learning engineers), and those that are very difficult or impossible to automate (such as plumbers). ## This Time Is (Probably) Different There remain some hyper-optimists who think that AI will create a balancing effect: since new technology in the past has tended to create more jobs than it destroyed, AI will do the same. I can't help but think that line of thought is delusional. This is the first time in history where we have a technology that can potentially create long-term joblessness for most of the population. That's not an abstract idea, it's a reality that could translate into human suffering on a scale we haven't seen before. Even in the case of the Luddites, the "rebalancing" of labor took a very long time—long enough that if they'd tried to wait it out they would have starved. It's a particularly dangerous situation because, within a capitalist society like the United States where there's little emphasis on social safety nets, the only means of survival for most is working. If you can't find a job because there just aren't enough jobs to go around, how do you survive? In previous eras, if someone was made redundant at their job, it was possible to reskill and, in the long-run, have a decent shot at success. The problem here is that reskilling, in most cases, involves doing something like going to college (more on that in the next section). Between getting laid off and landing that new job through reskilling, people need to do something to survive. In past eras, that meant jobs that people weren't necessarily proud of: fast food service, construction, janitorial, and so on. They weren't fun, but they kept the lights on while people took night classes and searched for new beginnings. With AI in the picture, many of those lower-end jobs are no longer available. Companies like McDonalds are already placing automated order taking and food preparation systems in their locations, so flipping burgers may not be an option if you lose your job. Likewise, Uber and many other platform-based gig economy companies are actively looking for ways to replace humans using their massive datasets—another avenue of survival closed. The labor markets are also shifting so fast now that, by the time you are deemed qualified for a new job through reskilling, it may be irrelevant or so low-paying that you will have wasted your time. Spending years developing competence may turn into an exercise in futility since the rapid changes in the economy will generate severely negative ROIs for that self-investment. Furthermore, what about people who are within sight of retirement age (let's use 10 years as a baseline)? Those people still want to support themselves, but if they lose their jobs they may not be able to find another one ever again—if only because most employers have a strong bias against older workers. ## Harsh Political Realities Will there be some form of Universal Basic Income (UBI) to support the millions who simply can't get or keep a job due to AI's influence on the economy? Or will large swathes of the population have to suffer serious consequences before anything changes? These are open questions, although I think the unfortunate truth is the second option. Anyone who thinks the government or the owners of capital will bail them out of this situation is deluded. There is close to zero (recent) historical precedent to support that idea, particularly in the age of extreme inequality we find ourselves in. Although it's possible people will band together and finally agree to take care of each other in a meaningful way (that's what a society is for, isn't it?), I remain skeptical. For now, it's safe to assume you're on your own. --- # Credential Value Shifts Source: https://www.aisurvival.org/credential-value-shifts/ Section: What This All Means *[Figure: A credential is worth something only while it is scarce. Flood the market with degree-holders and its value collapses, the same way it does for anything else. The rare, elite ones hold steady.]* This is a controversial opinion, but it's true: credentials often protect people who aren't that smart. Emphasizing credentials is one way to ensure that people who aren't geniuses still have a way to move up in the world. The top 1% of performers don't really need a piece of paper, because their work speaks for itself. But we can't run a society on that since, by definition, 99% of us wouldn't qualify. It also means that you can get a degree in something that doesn't require massive amounts of economically-relevant brainpower. Most people aren't built to be engineers, scientists or doctors, and many people who _could_ do that kind of work would hate it. So as it stands, you can still get ahead if you have a degree in something less lucrative but personally fulfilling. I think this is largely a good thing, although it can definitely be a net negative when stupid people who come from affluent families use Ivy League degrees as a way to monopolize powerful positions in society. And to be clear, I'm not saying everyone with a degree is stupid, merely that it isn't the indicator of intelligence or competence many take it to be. It's quite possible to be highly credentialed and stupid, but it's possible to not have any credentials and be an idiot as well. But a stupid person with a degree will, on average, do better in the job market than a smart person without one. The overall point is that this educational mechanism protects average people and has done so for quite some time. With the advent of AI, that may all change. The degrees with value will either be directly relevant to building and maintaining AI-centric systems, or in fields where AI simply can't replace humans. ## Intensifying Competition for Degrees That pool of qualifications will by default be quite small, and those looking to improve their position in society will no doubt recognize what degrees are valuable. We're already seeing cut-throat competition when it comes to academic success, imagine what will happen when all those hardcore, "if my kids don't go to Harvard I will kill myself" parents realize their children have even fewer pathways to success! Remember: businesses need fewer and fewer people to run now, which means just getting a specific type of credential may not be enough. If everyone has a degree in machine learning and companies just won't hire enough of those degree holders, then the value of that degree goes down by default. It seems likely that the value of all degrees is going to go down hard except in a few cases (I doubt an Ivy League degree will carry less weight, as it's more about connections and "prestige" than anything related to skills). Like the problem of job scarcity, this shift in credential value will hurt many people from the lower echelons of society who want to move up. And similar to the job situation, it's doubtful anyone is coming to save you—so other options must be explored. --- # A Bonanza for Capital Source: https://www.aisurvival.org/a-bonanza-for-capital/ Section: What This All Means Now that we know who the losers may be moving forward (pretty much anyone who works for a living), who are the winners? In short, people with enough resources to take risks. This is the problem most workers, if they aren't in highly-paid positions, face: they don't have the money to take risks. Without excess capital to plow into ventures that could eventually turn into income-producing assets, it's pretty difficult to move up the economic ladder. A worker could get extremely lucky and stumble on/create something that becomes valuable, which is unlikely. Or they could commit to a life of crime and just steal someone else's assets, which requires you to get a job on Wall Street. Barring those possibilities, workers don't have the means to utilize the primary way to make a lot of money in a capitalist economy, which is investing excess cash. They're stuck in the death spiral of lower wages, fewer opportunities and a worsening of their overall quality of life. ## The Best Time Ever for Business? But what's bad for workers tends to be good for those with capital to spare. If you look at everything that's been discussed until now through the eyes of capital, you see endless opportunity. If you don't need to hire as many (or any) people, that means you can test out new markets, new products and new services quickly and efficiently. When workers don't have the leverage to ensure they get paid enough, it's easy to find high-quality, low cost labor for those situations where you have no choice but to hire someone. Overall, this means if you have the resources to start and scale your own enterprise, you're living through perhaps the best time ever to do so. As long as you have money, you don't need credentials, nor do you need to concern yourself with how an employer treats you—you can do as you wish, as long as you continue to make money. With all the cheap, scalable tools at your disposal, you don't even necessarily need a _ton_ of money. All you need is enough to afford a quantity of experiments that you can hit statistical significance with, then either scale what works or kill what doesn't. When you hit something solid, just hire some cheap labor, automate it as much as you can, and move on to the next market or product line. The problem is that this is risky if you aren't already wealthy. A person with only enough money to run an experiment or two is not in a position to succeed, since failure will wipe them out. Wealthy people, on the other hand, can afford to fail over and over again in the pursuit of a winning business. Cost of experimentation is a big chunk of what separates the winners and losers in any capitalist economy. With AI, this cost will continue to go down—but not to zero. You will still need some kind of seed capital to get going, and if you're stuck at the bottom of the economic hierarchy without high-paying options, you're going to have a bad time. --- # AGI Isn't Necessary Source: https://www.aisurvival.org/agi-isnt-necessary/ Section: What This All Means Even if AGI remains years away (and despite the accelerating timelines, it's still speculative), the AI we already have is more than enough to cause serious economic disruption. However, the dramatic changes we're already seeing in the economy demonstrate that AI is not necessary for the situation to get dangerous for many people. When I say "dangerous," I don't mean AI will transform itself into a superweapon and physically kill you. I mean it is dangerous to your ability to pay your bills and keep a roof over your head. Automation in its current form, whether we're talking about the most basic of automated systems or sophisticated ML algorithms, is capable of making large swathes of the economy redundant. This is very bad news for workers, and excellent news for businesses. Whether it's a net positive or net negative depends on how we respond as a global society. A pattern that shows up more frequently now is the replacement of workers during periods of economic stress. Businesses that are normally making enough money to support large payrolls panic and start to not only lay off workers, but find ways to automate them. Once the crisis ends, instead of hiring again, they simply continue using automated solutions. ## An Ominous Example from the Energy Sector Back in 2016, there was a serious dip in oil and gas prices that created a sort of mini-recession in the oil business. The per barrel price plummeted, and energy companies started to lay off employees in droves. There had been a boom in oil and gas prices leading up to that, so it's expected that some of the excess spending on employees would get cut when the price went down. But there's a lot more to this story than you might expect. Take a look at this chart: *[Figure: Source]* The price of oil and gas is correlated to the number of rigs in operation, so you can use it as a proxy for prices. What's worth focusing on here is the fact that _employees were not rehired after prices recovered in 2017_. Why? Because the energy industry went on an automation spree to compensate for the lower revenues they were seeing across the board. Once they weren't bleeding money anymore, they realized they could just sail off into the sunset with their newly automated systems. If you want more details, I highly recommend [reading this article by Scott Santens](https://medium.com/basic-income/the-real-story-of-automation-beginning-with-one-simple-chart-8b95f9bad71b). **This shift didn't require AGI.** Remember: this happened in 2016! The machines involved are not even close to intelligent, they're literal robots who do tasks that are repetitive and dangerous to humans. In many ways this is a net benefit for humanity, as it means fewer people risk their lives for the sake of energy products. That societal benefit is lost on people who are out of work and unable to provide for themselves and their families. All they know is they can't make a living because machines were more cost-effective than they were. Without decent alternatives or some kind of government bailout, they're out of luck. Even worse, they're out of luck even when the economy improves and they expect to get their jobs back. ## A Larger Pattern This points to a larger pattern, one that indicates economic downturns will continue to create incentives for investment in automation. In previous eras of human history, businesses could only take this so far before they had to give in and hire people again. Economic recoveries meant that businesses and workers both prospered once conditions improved. That may not be the case moving forward. It looks more and more like a zero-sum game, where losing your job in a recession means you can't just wait it out on unemployment. You may _never_ get your job back, even when the company you worked for can afford to pay you again. They're more likely to capitalize on the savings of automation, spending that payroll money on optimizations and upgrades. For businesses, recessions and other more mild economic downturns represent incentive structures that could lead to massive gains. While it's easy to sit on your laurels when times are good, an ugly economy forces business owners to find more efficient solutions. Our current level of technology means they don't need AGI to accomplish this across some of the global economy's most important industries. ## Don't Wait The larger point here is simple: don't wait. If you think you're OK now, you have to realize that new waves of AI technology are showing up all the time. If you aren't preparing yourself now you are likely to run into serious problems. Even without the arrival of god-like AI, your livelihood may very well be a target for automation. This is true for both workers and business owners. Even if you have a successful enterprise, AI can represent an existential threat faster than you realize. It's best to assume, and prepare for, the worst, rather than hoping everything will stay the same. --- # What You Need to Do Source: https://www.aisurvival.org/what-you-need-to-do/ Section: Survival Strategy Most recommendations for what to do in this situation involve reskilling, or improving specific skillsets you already have. While I don't think this is quite _wrong_, I do think it's a misunderstanding of the issues at hand. For one, nobody knows what skills will be valuable going forward because the rate of change within the economy is so rapid. What's valuable now may not be worth much by the time you get a degree or finish a long certification program. Second, the entire idea that skills alone are what propel you forward is misguided in my opinion. From my own anecdotal experience, I've seen people with very few definable skills make it very far in life, and I've likewise seen incredibly skilled people fail over and over again. Yes, if you're the best ML engineer on the planet, you will be fine—but the number of people who even get the opportunity (which requires a whole lot of time, money and natural talent) to develop those kinds of skills to a world-class level is bound to be tiny. For everyone else, the solutions aren't so simple. ## Mixing Tangible Skills with Intangibles My personal belief is that the most valuable skillsets moving forward will be mixes of tangible skills and the intangible ability to combine things from different domains. This allows for personal differentiation and a chance to avoid getting turned into a nameless, shapeless cog in the AI-driven future. So, with all that being said, the rest of this section is dedicated towards pointing you in the right direction more than anything else. There aren't any easy answers here, but the concepts that follow will at least put you in a place to consider what your real options are. --- # Leverage AI Tools Source: https://www.aisurvival.org/leverage-ai-tools/ Section: Survival Strategy When I first wrote this book in 2017, this chapter didn't exist. Now it might be the most important one. Everything else in this guide is about mindset, strategy, and positioning. All of that still matters. But if I had to boil survival down to one concrete action, it would be this: **get good at using AI tools.** Not tomorrow. Now. ## Multipliers I've watched a single marketer with ChatGPT produce campaign material in three hours that used to take a small team two weeks. A friend of mine runs a consulting business where he uses AI to draft proposals, analyze financials, and write client reports. He fired his two part-time contractors. Not because he wanted to, but because the math stopped making sense. A person who knows how to wield these tools well can be five to ten times more productive than someone who doesn't. That number sounds absurd until you see it happen. Once you do, the implications hit hard: if your competitor figures this out and you don't, you're cooked. It's the same dynamic as bringing a calculator to a math competition where everyone else is doing long division by hand. Here's what makes this interesting from a survival standpoint. AI tools are cheap. Most of them are free or close to it. You don't need capital, connections, or credentials to access them. Skill is the bottleneck, and right now that skill is rare enough to be extremely valuable. For once, the little guy has a real edge if they're willing to learn. ## Most People Use AI Wrong Most people interact with AI the way they use Google: type in a vague question, skim the first response, move on. Doing that is like buying a Formula 1 car and only driving it to the grocery store. You're barely touching what these systems can do. Getting real value out of AI requires a different approach. You have to learn what to ask for, which means giving the system enough context and specificity that it can actually produce something useful. Garbage in, garbage out applies here more than almost anywhere else in technology. You also need to know when to ignore it. AI is excellent at first drafts, brainstorming, data analysis, and repetitive tasks. It is terrible at judgment calls, nuanced interpersonal situations, and anything where being confidently wrong has serious consequences. I've seen people blindly trust AI output on legal and financial questions and end up in a worse position than if they'd done nothing. Domain knowledge still matters because you need to be able to spot when the machine is feeding you polished nonsense. Iteration is where the real leverage lives. Almost nobody gets a great result on their first prompt. You work with it like a sparring partner: throw something out, see what comes back, adjust, go again. Treating AI like a vending machine (insert question, receive answer) will always produce mediocre results. Those who treat it like a back-and-forth with a very fast, very knowledgeable collaborator will pull ahead. ## Get Moving Right now, being competent with AI tools is a differentiator. Give it a few years and it will be table stakes, the baseline expectation for participation in the economy. If you start now you build a lead. If you wait, you'll be scrambling to catch up alongside everyone else. Pick one task you do regularly. Something involving writing, research, analysis, or data crunching. Do it with an AI tool and compare the result to your normal output. Notice where it's better, and where it falls short. Then try again with a different approach. Expand from there. You don't hand everything over to the machine. That would be stupid. What you do is figure out which parts of your work AI can handle at 80% quality in 5% of the time, then redirect your energy toward the parts where your human judgment actually matters. ## Opting Out Is Losing Some people refuse to use AI on principle. They have concerns about the environmental impact, the labor implications, the ethics of training data. I respect those concerns and share a few of them. But refusing to pick up a tool your competitors are using is not a principled stance. It's a losing strategy. You can advocate for responsible AI development while also learning to use what's available. In fact, the people who understand these tools best are the ones most equipped to push for better policy around them. Ignorance doesn't make you virtuous. It just makes you easier to replace. Every other chapter in this book talks about developing skills and mindsets that keep you relevant in an automated world. This chapter is simpler than all of them. Go use the tools. Get good at them. Stop waiting for permission. --- # Experiment Source: https://www.aisurvival.org/experiment/ Section: Survival Strategy Although you could argue about how stable the world has _ever_ been, we are no doubt living through an era of pure volatility. New trends seem to pop up and die off on a daily basis, and it feels impossible to keep up. The best way to handle that is to 1) acknowledge that you _can't_ keep up with that level of complexity, and 2) experiment on a regular basis. No matter where you are in your professional life, whether you're a struggling worker or a successful entrepreneur, you must try new things _all the time_. Intricate planning and preparation flat out doesn't make sense anymore, unless you're the owner of a gigantic corporation and you can afford those kind of lag times—even then, I'm skeptical it's a good idea. This is where being the little guy, the insurgent in the war of attrition we call modern capitalism, is a net benefit. While big companies are bogged down in their old, slow ways and unable to swing around fast enough to catch trends, smaller fish can gobble up opportunities. In many cases, big companies won't even bother if the opportunity is below a certain threshold. But that threshold might be $100 million per year, which means if you find something worth $5 million per year, you're set. ## Simple, But Difficult This is a simple principle, but it's difficult to execute. Most people are not geared for this at all, as they'd rather have simple, straightforward answers handed to them—that's why most people work for someone else. It's a basic algorithm that keeps life simple. With this oscillating landscape, that way of thinking doesn't make sense. You need to find your own answers, and the only way to do that is to take action and see where it leads you. If you can do that consistently, you'll have a real chance of coming out of the AI wave a winner. Not sure where to start? Just come up with a list of 10 ideas, all of which you can start within the next 24 hours. Remove anything from the list that requires more than a week to put out there. Don't worry about things like forming an LLC, making a logo or building a website. None of that matters, all that matters is putting something out into the world so you can get feedback. A high-quality model that I utilize on a regular basis can be found within [The Cult of Done Manifesto](https://medium.com/@bre/the-cult-of-done-manifesto-724ca1c2ff13). I recommend reading it and revisiting those ideas frequently. --- # Adapt or Die Source: https://www.aisurvival.org/adapt-or-die/ Section: Survival Strategy No matter what happens, it's clear that there's one tactic that will guarantee you end up as cannon fodder in this economic battlefield: staying put. The problem now is not so much the rate of change, but the rapidly changing rate of change. Volatility can go from insane to nonexistent, then back again, so fast that you have no idea what's going on. In a situation like this, it may feel impossible to adapt. In some sense, there's truth to that idea: you can't adapt to every little change that comes your way, nor should you try to. There's no way you can keep up with the daily shift in trends, but you can do a better job of spotting (and adapting to) the larger patterns around you. People just largely don't want to do it. It's a natural human impulse to resist change and continue to act as if the world will stay the same for the foreseeable future. Our brains much prefer this stable illusion over the unstable, chaotic reality we live within. Adaptation is also energy-intensive, and it's normal to resist anything that demands energy from your body and/or brain. Many people also just flat out don't have the energy to dedicate to adaptation after working multiple full-time jobs, raising children, and so on. Those people deserve more help than someone like me can provide—it's not appropriate to hold their lack of adaptation against them. For those who aren't drowning in their day-to-day lives, it's up to you to use your resources to understand and respond to your environment. As someone who has spent far too much time reading and writing in isolation, I can tell you right now that, above all else, you must make contact with the world. ## Go Outside This is what most people miss about the adaptation process: in a world dominated by data-hungry AI algorithms, going out into the world is a great way to develop an advantage. It's so easy to isolate yourself and think you can keyboard warrior your way to success online, but guess what? There are a billion other people on the planet who have the exact same idea. Do you have the drive and energy of a Pakistani teenager who lives with the specter of poverty hanging over them every day? Are you willing to do work for the same amount of money when you have a cost of living that's exponentially higher? These are the economic questions that will get even more intense as cheap overseas labor gets replaced by AI. The answer is to go places and do things where these dynamics don't matter. One of the best decisions you can make is to stop staring at a screen all the time and go build an in-person network. It's amazing what you can accomplish when you step outside, into a world where algorithms can't see the data you're gathering! For example, many people struggle to get jobs because they spend all their time applying on sites like LinkedIn. The problem is that these processes are highly automated, recruiters have hundreds of applicants to sort through, and the only people who make it through are the proverbial 1% of candidates. It's a far more effective method to meet people in person, ask lots of questions, and talk your way into a job or partnership. You might not even be the best person for the job—but that doesn't matter when you aren't being compared to a gigantic pool of candidates! If the CEO thinks you're a cool person, that can override many of the data-driven, algorithmic processes which might have hindered you before. This is just one example, but it's a form of adaptation. The job market has shifted dramatically, to the point where clicking "Apply" over and over again just won't cut it. An adaptive mindset looks at these kinds of problems and recognizes a separate way, which is often simpler and more effective. More than anything, don't just do what everyone else is doing—if an adaptation becomes too widespread, it's worth questioning its value. _Note: I have another book which goes much further into detail about adaptation and how to use it to solve problems in complex environments. It's called_ [_Strategic Complexity_](https://www.strategiccomplexity.com/)_, and if you've enjoyed this book, you'll benefit from that one as well._ --- # Control Your Attention Source: https://www.aisurvival.org/control-your-attention/ Section: Survival Strategy Attention is a precious resource, and the modern technological landscape is designed to drain you of that resource. Everything around you–especially the phone in your pocket–is now designed specifically to tap into your attentional bank and steal what's inside. Learning how to build and defend your attention is therefore a critical set of skills for anyone who wants to become automation-proof. Spending all your time enveloped in YouTube, Facebook and other information-rich platforms does not help you–on the contrary, it actively hurts you. Attention is a finite resource, and you need as much of it as you can muster if you're actively fighting against the onslaught of automation. Learning and adapting rely heavily on attention, and if you're constantly running it down because you're drowning in information that's coming at you all the time, you won't get much done. This is because you have a limited amount of energy available to your brain every day, and the supply gets more scarce the more you engage your attention. For example, multitasking is a dangerous myth because it's actually a constant engaging and reengaging process that rapidly depletes your brain's energy. Learning to avoid the multitasking trap is a skill all by itself and you need to learn it. I can feel it every time I decide to watch something on YouTube or check some other social media feed before I write. If I make that mistake (which is getting less common), I can feel my energy drain away, sucking away creative flow that I need to put forth my best work. This is social media doing what it's designed to do: keep you sedated and in place. The big platforms much prefer that you sit and watch, as it gives them data and advertising dollars in droves. Your attention gets packaged and sold like a bag of groceries, but at a much lower price. Why is this relevant to surviving AI? First, by limiting your exposure to attention-killing activities like social media scrolling, you don't hand over data to AI overlords. Secondly, keeping your attention on point means you can focus on experimentation, building, and generally _taking action_ instead of playing a passive role in life. If you can't swing this, all is lost. The machines have won and you will get sucked under the wave of automation. Attention is all you have, don't just give it away. ### Learning Materials *** #### Read These [**The Organized Mind: Thinking Straight in the Age of Information Overload**](http://amzn.to/2uXpcy1) **by Daniel Levitin** I've recommended this book more times than I can count, and that's because it's the best book on the market for anyone who wants to learn about how to manage their brain's resources. You have zero excuses not to read this book–it will change your life. [**The Attention Merchants: The Epic Scramble to Get Inside Our Heads**](http://amzn.to/2v0c3m1) **by Timothy Wu** Although this isn't a guide, per se, it is worth reading because it outlines the business motives behind harvesting attention on a large scale. You should know how your attention is turned into dollars, and this book is an excellent guide to the industries that are actively building their model around that dynamic. [**Hooked: How to Build Habit-Forming Products**](http://amzn.to/2urDljv) **by Nir Eyal** _Hooked_ is the playbook tech companies use to build products that capture your attention. Reading this will give you everything you need to identify when and how products are designed specifically for harvesting attention, which will in turn help you avoid those products. [**The World Beyond Your Head: On Becoming an Individual in an Age of Distraction**](http://amzn.to/2uXiBnd) **by Matthew B. Crawford** This is a bit philosophical, but it does contain some gems about how to manage your attention in a world filled with people trying to steal it from you. [**Autopilot: The Art and Science of Doing Nothing**](http://amzn.to/2hpbXRG) **by Andrew Smart** Autopilot is a book that explores the effects of being constantly bombarded by information and how you can mitigate the resulting damage. A large part of building your attention is taking time off from technology, and Autopilot is a great guide for learning how to do that. **Watch These** [**Advertising Hazards: Your Attention is a Commodity That Can Be Manipulated**](https://www.youtube.com/watch?v=uakWt0DTy7E) **by Big Think** Tim Wu, author of [The Attention Merchants](http://amzn.to/2v0c3m1), gives a synopsis of how attention is harvested and resold. [**Attention: Cognitive Psychology**](https://www.youtube.com/watch?v=bRej7Ak-QBA) **by Dr. Boaz Ben David** Can you define what "attention" means? If not (or even if you think you can), you should watch this brief video. [**Divided attention, selective attention, inattentional blindness, & change blindness**](https://www.youtube.com/watch?v=s4JBqLoY3tY) **by Khan Academy** A short overview of some key concepts from the world of attention research. ### How to Take Action Start to take control of your attention by finding ways to keep the world from stealing it. Here are some actions you can take right now to start taking back your attention: * Remove as many notifications as possible from apps on your phone. * Uninstall distracting social media apps from your phone. * Remove distracting apps from your home screen if you can't uninstall them. * Put your phone on silent when you need to focus. * Set a time every day for unplugging completely. Hobbies can also be used to build your ability to control your attentional resources. Here are some secondary activities you can start engaging in to develop your focus and build your attention: * Meditation * Playing a musical instrument * Martial arts (I recommend jiu-jitsu) * Learning a new language * Surfing * Picking up a new sport All of these are leisure activities, but they require you to put your attention to use and exercise some amount of discipline (there's that concept of transfer in action!). --- # Learn Persuasion Source: https://www.aisurvival.org/learn-persuasion/ Section: Survival Strategy Knowing how to get other people to do what you want is an invaluable skill. If you can get what you want with words (and following up with actions, of course), then you don't have to rely exclusively on what you're physically capable of doing. You can make friends, get jobs and generally progress in the world faster if you know how to get people to see things your way. Now, here's a nuance that matters: AI has gotten _very good_ at generating persuasive text. LLMs can write compelling sales copy, craft arguments, and handle analogies, metaphors, and irony with surprising skill. If persuasion were purely about words on a screen, AI would already be a serious threat to anyone in the business of persuading others. But persuasion between humans is about much more than words. It's about trust, rapport, body language, shared context, and the kind of relationship-building that happens over handshakes, lunches, and years of demonstrating reliability. AI has commoditized the _verbal and written_ components of persuasion, which means those skills alone are worth less than they used to be. What this actually does is make the _human_ elements of persuasion _more_ valuable, not less. When anyone can generate a slick sales email with AI, the differentiator becomes the person who shows up, reads the room, builds genuine relationships, and earns trust face-to-face. In-person persuasion hasn't been automated. Only the commodity layer of it has. It's also a skill you can take directly to the bank. If you can persuade, there are legions of businesses who will hire you to persuade people to purchase their products and services. Sales (which is business-oriented persuasion) is a profession with strong long-term prospects, especially for those who excel at the in-person, relationship-driven side of the work. Not only can you get a job as a salesperson, you could start a business and, using your powers of persuasion, get other people to hand money to you directly (as opposed to handing money to your employer). Every business requires sales expertise, and you can sell a well-developed persuasion ability for large sums of money. Even if you don't want to take on persuasion as a profession, having a silver tongue can help you get through job interviews and other screening processes. It might seem unfair, but the reality is that job skills on their own are rarely what determines who gets the job–the people with the strongest ability to sell themselves are the ones who get hired. Lots of people feel like sales and persuasion are "sleazy" or somehow dishonest. There are certainly elements of truth to those ideas, since persuasion is often used to rip people off in various ways. However, the reality is that persuasion is required for anyone who wants to survive in the world–regardless of whether you're in danger of being automated or not. If you can't get people to see things your way and help you achieve your goals, you're setting yourself up for failure. ### Learning Materials *** #### Read These [**Influence: The Psychology of Persuasion**](http://amzn.to/2ff2onH) **&** [**Pre-Suasion: A Revolutionary Way to Influence and Persuade**](http://amzn.to/2v4u8kM) **by Robert Cialdini** _Influence_ is the best possible starting point for anyone who wants to understand the basics of persuasion, and _Pre-Suasion_ builds on that framework by going into detail about focus and attention. Both are mandatory reading for anyone who wants to improve their persuasion skills. [**Popular: The Power of Likability in a Status-Obsessed World**](http://amzn.to/2hv9mFR) There are loads of books out there about how to be more likable, but this one was written by a psychology professor who actually knows what he's talking about. This book is filled with science and light on self-help fluff, which makes it much more valuable than other books in this genre. [**The Confidence Game: Why We Fall for It...Every Time**](http://amzn.to/2v5IIXj) **by Maria Konnikova** Con artists have always fascinated me because they're essentially in the business of selling nothing. They are so good at persuasion that they can get people to part with their money using only their words and a few minor deeds. Even though I find their trade completely unethical, there are lessons to be learned from people who do this. Even if you don't want to try to improve your own persuasion skills through their example, this book is helpful because it can help you spot when someone is trying to con you. [**The Ultimate Sales Letter: Attract New Customers. Boost your Sales.**](http://amzn.to/2v627aT) **by Dan Kennedy** Persuasion is not limited to speaking directly to other people. In fact, learning how to write in a persuasive way can be much more impactful–a persuasive piece of writing can be distributed all over the world quickly and efficiently, putting your message in front of millions of people. But writing in a way that's geared towards selling is not the same kind of writing we're taught in school, and learning to write this way requires a re-education of sorts. This book will get you started on the path towards writing that converts words into dollars. [**Payoff: The Hidden Logic That Shapes Our Motivations**](http://amzn.to/2weiEcZ) **by Dan Ariely** Dan Ariely has written a plethora of books about behavioral psychology, and they're all worth reading. However, this short guide is the most relevant for our purposes because it's tied directly to persuasion. It explores what drives people to do what they do, which you can use to your advantage whenever you're looking to utilize persuasion. [**Weaponized Lies: How to Think Critically in the Post-Truth Era**](http://amzn.to/2vpJocH) **by Daniel Levitin** _Weaponized Lies_ is an excellent guide for spotting lies and half-truths that are perpetrated in mass communication mediums. Much like the book about con artists, the value here is primarily in detecting when someone else is trying to persuade you with less-than-honest methods. [**The Art of Seduction**](http://amzn.to/2v6boj6) **by Robert Greene** Don't be fooled by the title: this isn't really about getting into people's pants (although that is definitely a form of persuasion). The concept of seduction here is more akin to putting people into a trance with your words. It's worth reading even if you're happily married and have no affair-related ambitions. #### Watch These [**The Science of Persuasion**](https://www.youtube.com/watch?v=cFdCzN7RYbw) **by Robert Cialdini** An animated introduction to the ideas that form the core of persuasion, as laid out in Cialdini's book _Influence_. Don't watch this _instead_ of reading the book, but use it to give yourself a quick primer _before_ reading it. [**The Science of Productivity and Motivation**](https://www.youtube.com/watch?v=xqMO\_gx8Wac) **by Big Think** Dan Ariely walks through the basics of motivation. Like the video above, this is not a substitute for reading his book, but is a good way to get introduced to the ideas. [**Lessons from the Best Salesman in the World**](https://www.youtube.com/watch?v=c0jdLhRCJRA) **by Joe Girard** This video is worth watching because you can observe the tonality, body language and overall demeanor of a world-class salesperson. The content of what he says is far less important than the way he delivers the message. #### How to Take Action If you really want to get better at persuasion, you'll need to find some goal that can only be accomplished by persuading another person to do something. Although reading and watching videos about persuasion will give you ideas about how to get better, you'll only make large leaps of progress by going out into the world and seeing how others respond to your approach. The simplest way to get started is to sell something. A good way to start is by putting something up for sale on a website like [Craigslist](http://craigslist.org). This will give you the opportunity to practice both your written (in the ad itself) and verbal (in-person interaction with the buyer) sales skills and make a little bit of money in the process. If you're feeling more ambitious, you can try selling a service of some kind. What kind of niche skills do you possess, and what would motivate someone to choose you over a competitor? How do you position yourself in terms of price and quality? Try picking up the phone and calling a business or person you think needs your service. It might fail (at which point you've lost nothing except a few moments of your time), or you might end up with a whole new stream of income. Want to put your skills to the test? Go get a full-time sales job. Sales is a tough job with a high turnover rate, but sales departments don't discriminate against people who don't have academic credentials or experience. If you can learn and demonstrate that you can close deals, you will have an income. --- # Act Strategically Source: https://www.aisurvival.org/act-strategically/ Section: Survival Strategy Strategy is a term that generally makes people think of armies clashing on a battlefield, but the reality is that strategy is part of everything we do. If you want to accomplish your goals in a variety of environments (in other words, be intelligent), you need to come up with strategies that you think will get you to those goals quickly and efficiently. It becomes especially important when other people are involved–they have their own goals and strategies for achieving them, and not having any methods for countering their actions puts you at a massive disadvantage. AI has gotten better at strategic analysis than most people realize. Modern AI systems can evaluate scenarios, weigh tradeoffs, and generate strategic options across a wide range of domains. If you're relying solely on your ability to analyze information and come up with plans, AI is gaining on you fast. But strategy in the real world is rarely a clean analytical exercise. It involves reading people, navigating ambiguous social dynamics, making judgment calls with incomplete information, and adapting on the fly when things go sideways. Humans still hold a decisive edge here, not because machines _can't_ process the information, but because the information itself is embedded in messy, real-world contexts that AI struggles to access. You have a brain that can gather data from conversations, body language, gut feelings, and random observations, then synthesize all of that into a coherent strategy. Developing this skill widens your advantage. AI can be a powerful tool for the _analytical_ side of strategy. The judgment, the intuition, the ability to act in ambiguous human environments? That's still yours. Embedded within strategy are lots of different sub-skills, such as creativity and leadership. I decided to bundle all of these under the single subject of strategy for the sake of simplicity. If you dive into these learning materials, you'll have everything you need to start not only being strategic, but also creative and able to lead other people. Like all the other skills listed in this book, the skill of strategy has a large amount of transfer. ### Learning Materials *** #### Read These [**The Joy of Game Theory: An Introduction to Strategic Thinking**](http://amzn.to/2v217WH) **by Presh Talwalkar** Start out with this short, light book, which walks you through the basics of game theory. Game theory is the formal study of strategy and involves quite a bit of math. But don't worry, this book won't go into that kind of detail–it's primarily concerned with explaining the key concepts in plain English. [**The Mind of War: John Boyd and American Security**](http://amzn.to/2v1duCg) **by Grant Hammond** On the surface, this looks like a book about one guy ([John Boyd](https://en.wikipedia.org/wiki/John\_Boyd\_\(military\_strategist), creator of the [OODA loop](https://en.wikipedia.org/wiki/OODA\_loop) concept) and his impact on the American military. The reality is that it's a deep exploration of strategy, history, psychology and sociology that will change the way you think about pretty much everything. [**The 33 Strategies of War**](http://amzn.to/2wqeT3j) **&** [**The 48 Laws of Power**](http://amzn.to/2uoGl02) **by Robert Greene** I included both of these books because they're both excellent distillations of strategic thought. Greene's books are always meticulously researched and filled with insight, no matter what the subject matter. You should go with _33 Strategies of War_ if you only want to read one for the sake of developing your strategic skillset. But I recommend you check them both out. [**Superforecasting: The Art and Science of Prediction**](http://amzn.to/2wqdDxh) **by Philip Tetlock** Learning how to evaluate information and assign probabilities to events is critical for forming quality strategic plans. _Superforecasting_ is the result of years of research done by Tetlock and his team on the nature of prediction, and it's filled with actionable information you can use to improve your decision making process. [**Strategy: A History**](http://amzn.to/2v21q3R) **by Sir Lawrence Freedman** You don't have to read this mammoth book in its entirety to get value out of it, but it is a magnificent volume. The first few chapters on their own are worth the price of admission for anyone who wants to be a better strategic thinker. #### Watch These [**How To Understand Your Enemies with Robert Greene**](https://www.youtube.com/watch?v=EDi4mxDW2cA) **by TheLipTV** Robert Greene, the author of several books about strategy (two of which are in the list above), walks through how Benjamin Franklin turned enemies into allies. Coalition-building is a key component of successful strategies and the lessons in this video are worth internalizing. [**Intro to Game Theory and the Dominant Strategy Equilibrium**](https://www.youtube.com/watch?v=3Y1WpytiHKE) **by The Economics Detective** You don't need to be an expert in game theory (it's a deep, mathematical discipline), but understanding the basic concepts–particularly the idea of dominance–will help you make better decisions. [**The Prisoner's Dilemma**](https://www.youtube.com/watch?v=t9Lo2fgxWHw) **by This Place** The prisoner's dilemma pops up all over the place, and explains the behavior of both individuals and companies in many situations. Being able to recognize it is of the utmost importance if you want to avoid lose-lose scenarios in life. Plus you can [win game shows](https://www.youtube.com/watch?v=S0qjK3TWZE8)! [**Offensive Strategy**](https://www.youtube.com/watch?v=r3NwaTXf1yg) **by Military History Visualized** This brief video reviews Sun Tzu's thoughts on offense. Even though it's usually better to find other solutions, sometimes you need to fight. Knowing how to fight then becomes a skill that you must develop. #### Play This Game [**The Evolution of Trust**](http://ncase.me/trust/) **by Nicky Case** This is a fantastic little illustration of how trust works in a strategic/game theoretic context. It's worth 30 minutes of your time to see how different strategies are affected by trust. #### How to Take Action If you want to get better at a skill, you need to practice. Strategy is no different. Use every situation you find yourself in as a testing ground for strategic thought. No matter how small, walk through each possible scenario you face and reframe your thoughts in a strategic mindset. Here are some ideas for how to do that: * Identify prisoner's dilemmas that you see out in the wild. I'm not going to give any of them away, but they're definitely out there. * Consider if a decision you're about to make is the dominant strategy or if you can do better. * Try to anticipate the reactions of those around you when you're trying to accomplish some goal. For example, think about how a loved one would react to a request of some kind and think about how you'd deal with all of those reactions. What strategies will absolutely not work? Which one strikes you as the dominant strategy? * Break down situations into game theoretic models. Consider if you are playing a simultaneous game, a sequential game, a non-cooperative game, etc. * If you're working a job, consider the different strategies you could use to get more assignments you want or a raise. How would you maneuver yourself around your coworkers and managers to get what you want? Make a habit of being strategic, and I guarantee you'll become a better strategist. And the better you are at strategy, the better your odds are for surviving automation. --- # Go Forth and Survive AI Source: https://www.aisurvival.org/go-forth-and-survive-ai/ Section: Conclusion This is the end of our journey together for now, but it's only the beginning of your journey towards being a more adaptable, versatile person. That squishy mass in your skull is capable of giving you everything you need to fight back against the tidal wave of automation that's taking the world economy by storm. What I've given you here is a starting point, a base of fundamental knowledge you can now use to build a better future for yourself and your family. The frameworks we've covered together will help you identify industries that won't be needing humans in the future and give you the versatility to pivot into a new career if you need (or want) to. I fully expect you to further develop these frameworks into something uniquely your own. There are all kinds of interesting niches out there that can't be filled by machines. If you stay flexible, keep learning and maintain a bias towards action, you can take advantage of opportunities you spot in your explorations of the world. Never forget: there is **always** a way forward. I also wrote another book about how to solve problems in complex environments called [Strategic Complexity](https://www.strategiccomplexity.com). It's a primer on how to handle the kinds of problems AI introduces into the world. Until next time: keep learning, stay flexible and keep your head up. \-Ace --- # Key concept: The Replaceability Test Source: https://www.aisurvival.org/concepts/replaceability-test/ From chapter: https://www.aisurvival.org/are-you-replaceable/ **The Replaceability Test is the heuristic I use for judging how likely a job is to be automated. Ask three questions about the work (Is it predictable? Does a framework for it already exist? Is data about it easy to collect?), then a fourth that can override the first three: how ambiguous is it?** ## The four questions 1. **Does the work involve predictable patterns of activity?** Very predictable work is a prime target for every kind of automation. 2. **Are there already frameworks for handling it?** Frameworks translate into structures that can be fed into an algorithm. 3. **How hard is it to gather data about what's being done?** Easy data means it may be easy to train a machine learning system. A quick proxy: count how many machines and pieces of software the job touches every day. Tech-heavy jobs leak data. 4. **How ambiguous is the work?** Even with piles of data, a job is hard to automate when it's full of human-induced ambiguity: lots of human-to-human interaction, built around systems that resist clean definitions. More "yes" answers to the first three, plus low ambiguity, means more exposure. ## What the test predicted When I first wrote this book in 2017, the framework flagged programming as a target. I caught flak for it, and most programmers were confident their jobs were safe. They were wrong. Software turned out to be one of the most data-rich, pattern-heavy professions in existence (open source code, documentation, Stack Overflow), and AI coding tools went from novelty to near-necessity in a few years. Other jobs, run through the same test: - **Copywriters:** already being automated. The job is drifting into copyediting machine output. - **Lawyers:** more exposed than I originally thought, and I used to put them in the "probably safe" category. Drafting and research are going fast. Counseling clients, negotiating and reading a jury are steeped in ambiguity and still human. - **Truckers:** eventually, but slower than everyone predicted. Driving looks mechanical and turns out to be deeply embodied (see Moravec's paradox). - **Police officers:** hard to automate. The job is split-second judgment in ambiguous, high-stakes human situations. ## Run it on yourself The chapter has an interactive version of the test. Answer five questions about your work and watch the meter score your replaceability, from defensible to prime target. ## Common questions ### How can I tell if AI will replace my job? Run it through the Replaceability Test. Jobs with predictable patterns, existing frameworks and easily collected data are the most exposed. Jobs heavy on ambiguity and human-to-human judgment are the hardest to automate. ### Which jobs are safest from AI? Jobs drenched in ambiguity and interpersonal judgment, and hands-on physical work in unpredictable settings, like plumbing or police work. Ambiguity protects a job far more than creativity does. ### Is there an online version of the test? Yes. The Are You Replaceable? chapter includes an interactive version that scores your job from defensible to prime target. --- # Key concept: Augmentation vs Displacement Source: https://www.aisurvival.org/concepts/augmentation-vs-displacement/ From chapter: https://www.aisurvival.org/are-you-replaceable/ **Augmentation vs displacement is how I describe what AI does to a profession: the same tools that augment the best performers displace everyone else. My go-to example is law. One lawyer with AI tools can do the work of three without them. A human still does the work, and two of those three humans are out of a job.** ## Both at once The debate usually gets framed as a choice: will AI help workers or replace them? It does both, often at the same moment, inside the same firm. Augmentation and displacement are two faces of one productivity jump. If one person can now produce what three used to, the company needs fewer people. So "AI is just a tool" offers less comfort than it sounds like it should. Tools that multiply output shrink headcount. ## Getting on the right side Survival means being the person who uses AI to multiply their value, not the person who gets multiplied out of a job. Here's what I recommend: - **Get good at AI tools now.** Someone who wields them well can be five to ten times more productive than someone who doesn't. The tools are cheap or free, so skill is the bottleneck, and for once the little guy has a real edge if they're willing to learn. - **Use AI like a sparring partner.** Throw something out, see what comes back, adjust, go again. Treat it like a vending machine and you'll get vending-machine results. - **Know when to ignore it.** AI is excellent at first drafts, brainstorming, data analysis and repetitive work. It's terrible at judgment calls and anywhere being confidently wrong is expensive. Domain knowledge is how you catch polished nonsense. - **Move before it's table stakes.** AI competence is a differentiator today. In a few years it'll be the baseline expectation for taking part in the economy. ## What it looks like in practice I've watched a single marketer with ChatGPT produce campaign material in three hours that used to take a small team two weeks. A friend of mine runs a consulting business and uses AI to draft proposals, analyze financials and write client reports. He let his two part-time contractors go. He didn't want to. The math stopped making sense. ## Common questions ### Will AI augment workers or replace them? Both, often at the same time. AI augments the top performers in a field while displacing the rest, because one augmented worker can do the work of several. ### How do I end up on the augmentation side? Get genuinely skilled with AI tools while that skill is still rare: give the system real context, iterate with it like a sparring partner, and learn to spot when its output is confidently wrong. --- # Key concept: Moravec's Paradox and Your Job Source: https://www.aisurvival.org/concepts/moravecs-paradox/ From chapter: https://www.aisurvival.org/are-you-replaceable/ **Moravec's paradox is the observation that machines are good at what humans find hard, like chess, math and code, and bad at what humans find easy, like walking, talking and grabbing things. It's the best explanation I know for why AI is hitting white-collar and creative work first, while plumbers and other hands-on trades stay hard to automate.** ## Why the paradox exists Walking, grasping and reading a room are ancient abilities, refined over the entire history of our species. Nobody has to teach you how to pick up a cup. Chess and programming are recent inventions. You have to be trained in them, and plenty of people find both hard. Those ancient skills turn out to be the hard ones to build machines around, largely because we can't even describe them to a computer. Decades and fortunes have gone into robots that walk like people, and they still haven't nailed it. ## What it means for work Everyone assumed blue-collar jobs would go first and creative jobs would be safest. It's happening the other way around. Designers and writers are losing work to machines fast. A plumber crawling under a sink, where every house has its own weird wrinkles, is far harder to replace. The same pattern shows up inside individual jobs: - **Fast food:** ordering is already automated with kiosks and AI drive-throughs. Food prep lags, because handling the thousand small physical variations of a real kitchen is harder than taking an order. - **Trucking:** the data is easy to collect, the task is well instrumented and the financial incentive is enormous. Autonomous trucks are *still* arriving slower than predicted, because driving is deeply embodied judgment. Information processing goes first; the physical parts of a job lag behind. Ambiguity is what protects work, and creativity on its own doesn't. ## Common questions ### What is Moravec's paradox? The observation, named after roboticist Hans Moravec, that tasks humans find easy (perception, movement, manipulating objects) are hard for machines, while tasks humans find hard (calculation, logic, chess) are comparatively easy for them. ### Why is AI automating creative jobs before physical ones? Because cognitive, information-processing work is easier to automate than embodied physical work in messy environments. Ambiguity protects a job far better than creativity does. --- # Key concept: Job Polarization Source: https://www.aisurvival.org/concepts/job-polarization/ From chapter: https://www.aisurvival.org/job-scarcity/ **Job polarization is the split I expect AI to force on the labor market: a small set of extremely well-paid jobs at the top and a large pool of low-paid, gig-style work at the bottom. Two kinds of work land in the well-paid group: jobs directly tied to AI (like machine learning engineers) and jobs that are very hard to automate (like plumbers).** ## How the squeeze works Businesses always have an incentive to shrink payroll, and AI hands them better ways to do it. One person can already start and run a successful business using cheap tools and contracted labor. Expect companies to keep finding ways around full-time hires, and the gig economy (selling labor in small, cheap bursts on platforms like Uber and Instacart) to keep growing. Software engineering shows the split already. Elite engineers, especially those building AI systems, command enormous salaries. Mid-level developers who mostly write routine code are getting squeezed hard. ## Why this time is (probably) different Optimists point out that new technology has always created more jobs than it destroyed. I can't help but think that line of thought is delusional this time, for four reasons: - **The fallback jobs are going too.** Laid-off workers used to survive on fast food, construction or janitorial work while they retrained. Automated ordering and food prep are closing that door, and gig platforms are looking to replace their humans using their own massive datasets. - **Reskilling is too slow.** Labor markets shift so fast that by the time you're qualified for a new job, it may be irrelevant or badly paid. - **Older workers get stranded.** Someone within ten years of retirement who loses their job may never find another, if only because of employers' bias against older workers. - **Nobody is coming to save you.** Universal basic income is an open question, and my bet is that large parts of the population will suffer before anything changes. For now, assume you're on your own. ## Where to stand Aim for one of the two protected ends: work tied to building and using AI, or work too ambiguous or physical to automate. The middle is where the squeeze is. ## Common questions ### What is job polarization? A labor market splitting into high-paid jobs at the top and low-paid jobs at the bottom while the middle hollows out. AI accelerates it. ### Which jobs will stay well paid as AI spreads? Jobs directly tied to AI, such as machine learning engineers, and jobs that are very hard to automate, such as plumbing and other skilled physical trades. --- # Key concept: Recession-Driven Automation Source: https://www.aisurvival.org/concepts/recession-automation/ From chapter: https://www.aisurvival.org/agi-isnt-necessary/ **Recession-driven automation is a pattern I keep coming back to: during an economic downturn, companies lay workers off and automate their jobs, and when the economy recovers they keep the machines and never rehire. It doesn't take superintelligence. Ordinary automation is enough.** ## The oil rig lesson In 2016, oil and gas prices crashed into a mini-recession, and energy companies laid off workers in droves. Some cuts were expected after a boom. The surprise came later: when prices recovered in 2017, the rigs came back and the jobs didn't. The industry had spent the downturn on an automation spree, and once the bleeding stopped it had no reason to go back. (Scott Santens tells the full story, chart included, in [The Real Story of Automation Beginning with One Simple Chart](https://medium.com/basic-income/the-real-story-of-automation-beginning-with-one-simple-chart-8b95f9bad71b).) None of the machines involved were remotely intelligent. They were robots doing repetitive, dangerous work. That's arguably good for society, since fewer people risk their lives for energy products. It means nothing to the workers who can no longer provide for their families. ## Why it only goes one way Recoveries used to force businesses to hire again, and owners and workers prospered together once conditions improved. That link is breaking. Every downturn now gives owners a reason to find efficiencies, and current technology gives them the means. A job lost in a recession may be gone for good, even after the company can afford to pay you again. Your old salary goes into upgrades and optimizations instead. ## What to do with this Don't wait. If you feel safe now, remember that new waves of AI keep arriving, and the next downturn is when the cuts happen. The warning covers business owners too: AI can become an existential threat to a successful company faster than you'd think. Assume the worst and prepare for it, instead of hoping everything stays the same. ## Common questions ### Do companies rehire workers after automating during a recession? Often not. In the 2016 oil downturn, energy companies automated during the slump and didn't rehire when prices recovered in 2017. ### Does AI need to reach AGI to cause mass job losses? No. Ordinary automation, deployed during economic downturns, is already enough to remove jobs permanently. --- # Key concept: Credential Decay Source: https://www.aisurvival.org/concepts/credential-decay/ From chapter: https://www.aisurvival.org/credential-value-shifts/ **Credential decay is the shift I expect in what degrees are worth: as AI lets businesses run with fewer people, most credentials lose their power in the job market. Two kinds should keep their value: degrees directly tied to building and maintaining AI systems, and degrees in fields where AI simply can't replace people. Ivy League degrees, which trade on connections and prestige more than skills, will probably hold steady too.** ## What credentials have been doing This is a controversial opinion, but it's true: credentials often protect people who aren't that smart. The top 1% don't need a piece of paper, since their work speaks for itself, but a society can't run on the top 1%. Degrees give average people a way up, including people who pick fulfilling fields that don't pay much. On average, a less capable person with a degree still beats a smarter person without one in the job market. That protection is what AI threatens. ## Why the value drops - **Fewer jobs per degree.** If everyone holds a machine learning degree and companies won't hire enough of them, the degree is worth less by default. A credential only has value while it's scarce. - **More competition for the few that still work.** The pool of valuable qualifications will be small, and ambitious families will pile into it. Academic competition is already cutthroat. Picture the "if my kids don't go to Harvard I will kill myself" parents discovering there are even fewer paths to success. - **The people at the bottom lose most.** Credential decay hits hardest the people from the lower rungs of society trying to climb, the same people hit hardest by job scarcity. ## What to do instead Don't count on a credential to carry you, and don't expect anyone to fix this for you. My survival strategy leans on things no degree certifies: skill with AI tools, constant experimentation, adaptability, in-person networks, persuasion and strategic thinking. ## Common questions ### Will AI make college degrees worthless? Not all of them. Most degrees will lose value as companies need fewer people. Degrees tied to AI, degrees in fields AI can't replace, and elite prestige degrees will hold up best. ### Why would AI reduce the value of degrees? A credential is valuable while it's scarce relative to the jobs it unlocks. When AI shrinks the number of jobs, the same degree buys less, and competition for the few valuable ones intensifies.