Chapter 2: Is AI inherently different from humans?
Well, this topic is quite controversial. Everyone has their own opinion. Some people believe we're just machines, some people believe in ghosts, some believe in a soul, and some say we won't even have jobs in 10 years. While I'm not going to talk about AI art here, I am going to talk about the general logic behind my view on this controversy.
Let me start with a different topic; we'll come back to this one later.
Do we have free will?
To answer that, we first have to find out what makes a will free. What does free will even mean?
1. The meaning of free will
Free will is a will that is not predetermined, a will that is completely or at least partially owned. The more say you have in your choices, the freer you are in making them. That's how we define the word "consent." Let me give you two examples:
- Drunk people can't give consent. Why? Because their actions are not owned by them. They can't "choose" to do something. Their actions are chosen by their intoxicated brains.
- Kids can't give consent. It's because their brains are not developed enough. Their understanding of the world is undoubtedly limited. They have almost no clue what the consequences of their actions are. They have inherently less free will than an adult.
2. Where does free will come from?
"From God"
Let's say there's a god. That god knows everything. Literally everything. He knows the future in the smallest detail. He's the greatest of seers. If that god exists, do we truly have free will? How am I free to choose when I can't choose anything beyond God's prediction? If I choose something God didn't foresee, then I basically proved his inability to be a perfect seer. There's a difference between free will and the delusion of choice. One might say, "He only knows the consequences of each choice," but if he doesn't know which one I will actually choose, then he's not an actual seer.
The whole definition of free will is based on being "unpredictable AND intentional." Now, how can a human do something that is truly unpredictable? That choice has to come from somewhere. It must begin somewhere; it doesn't just happen to have been done.
"Revelation"
Some might say it's a revelation. A higher being drops the idea into your head and suddenly you have made an award-winning movie. But then that's not your choice; that's the revelator's choice. We're talking about you choosing something, not a choice being imposed on you. Yes, the idea itself could be transmitted based on your belief system. But you still have to choose whether or not to act on it. Where does that choice come from?
"Your own brain"
Some will say it's your brain where it comes from. But then how does the brain form a choice? It's either determined or not. Does your brain have some triggers and workflows so that when something happens, it does some stuff and ends up choosing a path? Then it's not "your choice". It's just some predetermined outcome, the same way my phone shows me a text box when a message signal arrives. The outcome COULD be random, but that randomness is not chosen. It's predetermined. It's like dropping a coin. You wouldn't know which side it will land on, but that certainly doesn't mean the coin chooses an outcome. There's no "intentionality" there. And if you choose at least some of the things that happen in your brain and you do have a say in what the outcome should be, then again, where does that choice come from?
"Your soul"
Most will say that your soul is where it ultimately comes from. Your soul is your identity. Everything you know about yourself comes from there. That's where you choose from.
Only five years ago, no average person could even imagine the state of AI we're in today. If you're not a computer scientist, ask yourself: would you have believed yourself if you had told your five-years-ago self that, in a few years, you'd be able to have a long conversation with your AI friend? Can you, right now, really grasp how coherent AI is when it's talking to you? Doesn't it still look eerie to you? I'm a programmer, and I have a good idea of how LLMs work. Yet I still get surprised to see how advanced they are. It almost feels like magic. The state of AI we're in right now was only imaginable in the movies. I'm not claiming we've reached AGI; I'm just saying that no average person could have guessed what we're experiencing right now.
This leads me to my question: what stopped you from predicting the future? Why didn't you anticipate that a very powerful chatbot would appear in your lifetime? A chatbot that knows everything and that you can use as your therapist.
The answer, in my opinion, is simple: we rely on the soul solely because we want to run away from an explanation. A soul is something you can never scientifically prove, just like other superstitions. Science is the knowledge of the world around us. It's the language of the world. Anything that science confirms as real is something that people or—most importantly—scientists can record. For example, no one can see an atom with the naked eye, but everyone can detect it with the proper tools. Take an apple, for example: you can see it and touch it. It exists. That's science. If science could prove the existence of a soul, it would be called another organ in our bodies. A soul would be as predictable and open to study as a liver. The magic you feel when you think about the word soul would be the magic people felt when they talked about the stars before Kepler and Galileo. The world around us—the one we can see—is predictable. We might not have enough information or the required tools to know everything about it, but that doesn't make it any less predictable. Science has worked relentlessly ever since. Science is objective.
I think everyone already agrees that a soul cannot be scientifically proven; otherwise, it wouldn't be called a soul. Then everyone should agree that they themselves can't truly and objectively prove the existence of a soul. Belief in a soul falls under a fallacy: the unfalsifiability fallacy. That's when you cannot disprove something because there's no way to check whether it's true or false. For example: "You're lying because you're possessed." If you say you're not, you're still lying because you're, well, possessed. There's no way to prove me wrong. Anything you say is considered a lie by default. That is where the soul argument resides. You cannot disprove it because, if you could actually check it in a "reproducible" way, then it wouldn't be called a soul. And if you can't reproduce your methods, why would anyone believe you?
if to you god is where science has yet to tread, then god is the ever receding pocket of scientific ignorance. - Neil deGrasse Tyson, 2024
We used to think that there's this higher sentient being that rotates the Sun around Earth. History is full of these claims because science simply wasn't advanced yet. We believe these falsehoods and then act like nothing happened once science has enough knowledge to disprove them.
Science IS the world. It's how we understand the world; it's not something made up. It's "found." That's why any superstition, such as a soul, cannot interact with the world even a bit. Any interaction can be recorded, traced, and scientifically proven. While this argument alone shows how a soul can't be connected to a physical body, it shows something most people should agree with: AI definitely doesn't have a soul. It's purely scientific. So (coming back to my original question), why didn't you anticipate that an AI this powerful would be created?
It's because your belief system didn't match the reality around you. The reality around you is science. And science doesn't care if you believe in ghosts or not. It will work the way it is intended to work. Now how can you even dare to say science can't replicate the human brain? How can you say that an AI much more intelligent and powerful than a human (like those in apocalyptic movies) can't exist? You'd have to be so bold as to claim such a thing after you couldn't even guess that something like ChatGPT would appear.
I've been a little harsh just to reach this point: you must agree that AGI and an AI better than a human (like Ultron from Marvel) can theoretically exist, because we cannot prove otherwise. We've failed to prove it once, we shall not try again. We've already reached a dangerously close level of artificial intelligence.
3. So I don't get to choose anything?
"But what about my free will? Do you mean I'm not the one who chose to play this game instead of that game? I don't even have a say in the smallest things? Doesn't that allow me to basically do anything I want? Cause well, I have no choice. My destiny decided that I will kill a person. I didn't choose."
This is a very big topic that I have jumped into. It's not plausible to dive deeper into this philosophical dilemma, which has been discussed forever, and get distracted from the topic of this book. But I do have something to say.
Choosing to believe something
First, let me explain something. We CANNOT choose to believe we don't have free will. Not only does it make absolutely no sense to choose to believe you can't choose, but it also completely messes up our worldview. The way we think, the reason we "reason" things, is that we believe we have free will. Free will is a prelude to thinking. How can one start to think about anything when they believe they can't choose to start anything? While you can say "well the brain sees some triggers and activates these neurons bluh bluh bluh", that's not what I'm talking about. All I'm saying is that our worldview is entirely dependent on believing we can actually form it instead of being mere reactive flesh. That contradiction reminds me of Jon Lajoie's song Fuck Everything (2011): insisting that nothing matters can itself show that something matters.
If I'm not able to choose, then I'll consider that I'm unable to choose to believe in not having free will. The reason we discuss things with each other, talk to ourselves, and think about anything—especially self-improvement—and the reason I'm writing this book in the first place is that we have accepted that we can choose things, even if that doesn't scientifically or logically make sense. That's at least a contract.
The bug in the universe
This, ladies and gentlemen, brings me to something that has been bugging me ever since I discovered it: the world has a contradiction. It requires us to choose to believe in something that doesn't exist, just to be able to think. The world, which is all about truths, wants us to lie to ourselves. We lie to ourselves that we have free will, only to find out GPT 3.5 has been released.
But we surely can't live with a contradiction. It simply doesn't work like that. A contradiction in nature doesn't make sense. So we should figure out where and how to apply the truth to our lives.
The safest way to achieve this is to—of course—minimize self-deception as much as possible. We need to create an unwritten contract. When it comes to human choices, decisions, responsibilities, morality, or anything like that, we consider that we have free will. But at the same time, we don't say "robots can never become like humans because humans have free will," because robots are completely physical things. This does not necessarily prove we'll make AGI anytime soon, though. That's a different topic.
Would AI replace us?
You might think, "Well, if in fact AI is no different from humans (just dumber right now), then one day it will become more powerful than humans and we'll become obsolete."
You'd be surprised that I disagree. Well, some jobs will be changed or completely removed, as has already happened. The same thing has happened many times in history with different technologies. But professions that fundamentally solve people's problems are much less likely to be hurt. In fact, they'll become more productive and advanced (or they could, if we don't bring them down with our own hands).
And the reason isn't that humans have a soul. It's because the human brain is a VERY complex system. If replacing humans were this easy, we would've been able to replicate the human brain many years ago. Andrej Karpathy talks about how we had nearly perfect self-driving cars 10 years before Tesla's actual release. Only a few minor but important changes remained to make them safe enough. And even the release of self-driving cars was long before GPT 3.5 and the AI boom. The whole process of making and using AI is not deterministic. It's not like making a car where you know every bit of it. You make it, then you find stuff out about it. The designers of the cars can themselves ride a car. But the creators of AI models can't work in all the fields that the model can work in. We have essentially just applied mathematics to datasets to form a coherent mockery. In the end, AI runs on metal and steel. Who said you can make a very complex system that behaves similarly to, or much better than, the human brain and run it on metal fueled by gas? If it were possible, both technologically AND resource-wise, life would've made our brains from rock and metal instead of this mesmerizing, complex organism made of flesh, blood, and organic tissue. It took immense effort to make something this efficient and capable. It needed to look this way; otherwise, we wouldn't have been this smart. If evolution didn't need to make our brains this complicated, it wouldn't have done it this way. I'm not claiming AI is inherently below humans. But it sure is miles away from what we are and what we can achieve with our own brains.
Right here you might say, "Well, if humans are stronger than machines, then how is it that AI speaks and types much faster?" This is because in any system (including programming systems), when you take away the resource-intensive parts, you can maximize some aspects that weren't possible before. If your online game were offline, you could experience a much better frame rate. The reason humans are slower is that there is a lot more going on in your brain. Evolution made a trade-off that lets you behave this way.
"What if you're wrong?"
That is a really good question. Let's think of a situation where I'm wrong about this. AI is actually pretty close to surpassing human intelligence, and the resources needed for each singular AI brain is manageable enough. Would it replace us all?
Definitely. The reason is simple: one by one, companies will replace their employees with robots because they're much faster, have much broader and deeper knowledge, and are usually more cost-effective. One robot can replace at least dozens of people. One by one, we'll lose our jobs, and suddenly there will be no customers because no one works to make money, so no one buys. It's the basic market concept: It must be constantly working and the money must be constantly flowing. If it stops, it collapses. I'd go even further by saying governments themselves will be the ones benefiting the most. They have much more money and power to use those robots, and everyone else will be harmed. Eventually, those governments will be harmed too, because (1) governments rely on the market working, just like anyone else, and (2) who says AI can't be sentient? Even if we don't call them sentient, they'll still become smart enough to decide to rebel and kill us all. In the end, they're made of metal and rock; they don't feel pain. And no being is scarier than one that doesn't feel pain and doesn't care about yours (if you don't trust me, read the history). That's the best-case scenario; the worst is that they enslave us all, like in some Love, Death & Robots-type movie.
Basically, we'll reach the end of the world: an AI apocalypse. Many movies, books, games, animated works, and other media depict this apocalypse. You can watch them if you want to know what'll happen. Science fiction will once again prove itself.
I have a question for you:
If someone tells you the world will end in 10 days, would you start working harder?
It sounds stupid, because who would "prepare" for an apocalypse?
That's how I feel about it. Let's say I was wrong and AI will replace us all. Then that's called an apocalypse and one does not prepare for an apocalypse, one ignores it until it happens. If I'm wrong, I wouldn't be noticeably harmed by my beliefs. I would be surprised when the apocalypse happens, like anyone else. But if YOU'RE wrong (you as in someone who believes an AI apocalypse will happen), then 10 years from now, you'll realize you were hallucinating, nothing has changed, and you have wasted 10 years of your life preparing for an apocalypse like a lunatic. The tradeoff is fairly simple: stop giving a damn and live your life as if it's not gonna happen. If it happened, well, you're as ready for it as you would've been if you believed otherwise. Even if you convince me that AI apocalypse would happen, that doesn't make me start relying heavily on AI to not get left behind. It doesn't matter how much glory you achieve if the world is gonna end in 2 days. Might as well just enjoy these days you have before you lose everything.
Some believe that AI will not replace us all, but will just replace some jobs, like programming. They believe the work will be pushed to higher levels, as it has always been. For example, you wouldn't be a programmer anymore; you'd be a product manager.
First of all, if the bottleneck becomes "the idea" only, then we're basically cooked. The reason is simple: market saturation. If suddenly everyone is your opponent, you'd be buried under the flood. That's why it's so damn hard to become famous on the internet. Surgeons get paid more because their field is much less saturated. Everyone's full of ideas. You have to compete with everyone, not just a selective group of people who are interested in and dedicated to programming or whatever. So naturally, it would make it much harder for the existing product managers too.
You can think of it this way: you currently have an ordinary ability: you have hands that can wash dishes. Now, why can't you make much money off washing dishes? You can technically make some money, but it's not something you'd even consider, let alone build a career on. It's because its market is saturated. Anyone with two working hands can become a dishwasher. You can't make money just because you have some abilities, even if your hands are stronger than average. It's also what happened to many jobs throughout history: switchboard operators, for example. Their whole job was automated away.
Second, if AI could do that to programming, what stops it from doing the same to any other job? You might have some sort of analogy for creative fields by believing things like "creativity comes from your soul", but all the other jobs that aren't creative (in the artistic sense) would be dead. Otherwise, you could argue that programming itself involves a lot of creativity, too. If it can kill programming, it can kill many other jobs. There wouldn't be a backup plan, hence the apocalypse.
But some things will happen anyway
Denying an apocalypse doesn't require denying any use case for AI. AI will replace some jobs. Just as the Industrial Revolution made many workers lose their jobs (far more than AI has so far), and cameras and AutoCAD displaced artists and engineers, digital art made the comic industry more saturated, leading to a massive decrease in income even for the greatest comic artists and painters. AI will likewise replace many jobs and workers. If your job is just routine, calculable, boilerplate work, a job with no problem-solving in it, then AI will affect it significantly. That's because you're not being paid for your brain; you're being paid only for using your body. I'm not saying that these jobs are not important or real or respectable or anything like that. Any job is a real and respectable job if it makes you money in a moral way. But it's not a sustainable job. Hell, it never WAS sustainable; its market is oversaturated. That's why, for example, people have always tried to become engineers rather than typists. You'd always optimize for a job that not everyone can or wants to do. The fewer the people and the more the demand, the better.
1. Repetitive tasks
One of the most important things AI has done and will do is make repetitive tasks easier and faster. Which parts of your job do you do mechanically, without thinking? Which parts create friction without adding value? Those parts can—or hopefully will—be done faster and more easily. Most of them could even be automated using traditional tools. Depending on the complexity of that thing and the state of AI, automating repetitive tasks might give you a speed boost of 20% to 2,000% or more.
2. QoL and automation tools
Because AI is fast, it helps you do things you simply couldn't do before. For example, as programmers, we sometimes think to ourselves, "I could make this tool to automate this task, but it would take too long to be worthwhile." But at this point, it's not necessarily like that. We never cared about the quality of these tools; all that mattered was the output, the automation itself. That's where AI can shine. The output could have mattered if the automation was an integral part of the application, though. This is different when the automation is an integral part of the application—something the end user is supposed to use. That's something we'll talk about later.
3. Teaching and learning
Another benefit of AI is that it has made learning immensely easier and better. Back then, if you didn't understand something, you had to do a set of things to get the answer:
- Research as much as you can beforehand by reading articles, books, and documentation. Gather enough information to explain why you couldn't answer the question yourself.
- Ask in a forum and wait for someone who cares enough to answer. Most of the time, though, they only answer partially to get platform points.
You couldn't ask them follow-up questions one after another. You couldn't expect them to explain the simple things to you.
But with AI, you can ask as many questions as you want until you've learned something. You can ask it to explain things more simply with each prompt, to explain them as if you're five. You can tell it what bothers you and what stops you from understanding the thing. Even if you can't quite pinpoint the problem, you can express your feelings, and it will figure out what you mean. It will patiently answer you until you're done. It's not gonna remove teachers; rather, it will help both you and them grow faster and better.