This post is going to be largely a rehashing of Adam Conover’s recent video on the topic, but I already held these opinions before seeing that video. If you’d rather just watch that, here you go:

Immediately after watching that video, I saw that our dear friend Hank Green made a video about how terrible AI is going to be for humanity:

It pains me to say it, but Hank is just wrong. Adam is much closer to being right (if not just straight up completely right).

AI is a hype tool made up by tech bros to inflate market valuation. A few years ago it was “crowdsourcing”, then “the internet of things,” and then it was “the metaverse.” Next came self-driving cars, and then it was web3 / crypto. Now it’s AI. What’s really at play here is that all the of the easy gains from the invention of the microchip have already been made, but tech bros don’t wanna believe that the ’90s are over and that the smart phone has already been invented. They wanna believe that the potential for microchips and the profits to be gleaned from them are infinite (profits that in no way benefit the working class, I might add).

Just like crypto is a cute trick that can technically work, and the metaverse actually DID work when it was called World of Warcraft, AI DOES do SOMETHING. It can generate text, or images, or other kinds structured data based on pattern matching. That’s cool. That is useful. Is it going to change everything? Absolutely not. It’ll automate some tasks that rely on pattern matching that people can figure out how to turn into a training set (notice how self-driving cars is not one of those things).

This is what I DON’T see happening though: creative jobs getting replaced. If you try to replace creativity with an AI model, it’s because you’re incapable of recognizing creativity. AI is fundamentally incapable of truly unique creation, because the way it works is that it picks the MOST LIKELY next data point (word, pixel color, database row value, etc) based on previous ones. Creativity is, in many ways, what happens when the most likely next data point isn’t useful, so you have to find something different, or something that seemed completely disconnected and see how introducing that would effect what you’re working on. AI cannot do this. Sure, you can inject some creativity from the beginning based on the prompts that you write, but from that point on (the point after the human is involved) nothing creative happens. It’s just generating the most likely next data point. It’s not magic, and in the end it won’t be that interesting.

Have you noticed how most of what ChatGPT makes sounds like corporate word salad? Well, that’s because corporations are about as creative as AI is. They intentionally limit human minds to follow the same kind of thinking as AI in an attempt to pursue profits, and in the end ultimately become stagnant and uncreative places (even if the initial founders, motivated by true creativity actually did do something useful at some point). Maybe this is why certain people find AI so compelling: because it sounds exactly like what they themselves have been trained to think is “right speak.” Personally I find it pretty useless.

In the end AI models will reach their limits for what they’re useful for, just like everything else that’s been invented. They won’t be able to replace humans, because they’re not humans. You will never feel validated by an AI the way you are emotionally validated by a human, and an AI won’t ever feel your hunger or desires. Even if there comes a day where machines can do all the human things better than humans can, we won’t actually value it because it would be like if a whale told you about its day: maybe kind of novel for a while, but ultimately not that compelling because we are different life forms with different values driven by different needs.

And that’s assuming a “best case scenario” where a GPI is actually invented, and this isn’t all just techbro hype to drive market valuation. Which it is.

I’ll leave you with a link to Noam Chomsky’s critique of current AI technologies, which basically says all this in a much more eloquent way.