30 Things I've Learned About AI

In 8 years of studying and writing about it

1. To know if an AI tool works, don’t read news headlines—try it yourself

2. Even if an AI bubble explodes, the survivors will build a new technological landscape; and as a second-order effect, a new sociocultural landscape as well

3. AI won’t take your job, a person using AI will; most likely you using AI will replace yourself not using it

4. Companies care about shareholders, politicians care about votes, journalists care about paychecks, bloggers care about views, and researchers care about citations. Find the right mix of sources for what you care about

5. Generative AI is in its final stage, what comes next isn’t valuable for what it generates but for what it doesn’t

6. AI isn’t the technology “of the future” but of the past and the present; we just don’t call it AI once we use it in everyday life

7. AI requires regulation, but not at the expense of innovation

8. Some people saw coming a decade ago what’s happening today; follow them and you’ll see (part of) the future

9. In AI, everything (even the name), is and isn’t marketing at the same time

10. Those who over-hype in the extreme and those who anti-hype in the extreme are often cut from the same cloth

11. In times of turmoil and controversy, listen to the quiet ones; the rest are picking sides to build their identity

12. Studying the underlying math of neural networks gives you insights you can’t acquire elsewhere

13. Unless you’re good at making predictions, actual results from the field will catch you off guard one year from now (even months from now)—reflect on your surprise

14. AI won’t always be this cheap—this period was a welcome anomaly

15. Seriously trying out AI tools has a negligible cost to you but an incredible potential upside; the only people who purposefully don’t do it are those too invested in discrediting—and even tearing down—the AI edifice

16. GPT-5 will be surprisingly good; you can’t anticipate technology that’s being invented as we speak

17. You’re not the audience of hype; AI is valuable but also expensive to build, that's why companies hype it, to entice investors so they keep the market alive

18. Generative AI is a small part of AI, most of it is predictive (e.g. spam filters, identification systems, social media filters, and forecasting)

19. In ten years, most (if not all) writers will have incorporated AI tools in their workflows

20. AI has become like politics; people have chosen their sides, and no amount of new information will change their stance, as they’ve locked themselves into a tribal identity instead of staying open to updates

21. Knowing a lot of AI ≠ being rich overnight

22. I believe AI x-riskers (the “AI may kill us” crowd) are, for the most part, honest people despite their unusual views and the intensity with which they defend them

23. I believe e/accs (the “accelerate or die” crowd) are in the game mostly for their own interests, despite I kind of agree with techno-optimists more than x-riskers

24. Sam Altman doesn’t care about money but about legacy—not sure what’s better

25. You can massage AI data, statistics, and their interpretations to make them favor any argument, from black to white—beware the man of one study

26. China will win the AI race if the US doesn’t do a 180-degree mindset shift; individual freedom often comes at the cost of social cohesion, which hinders adequate action

27. A few people love AI; a few more hate it; the vast majority are either indifferent or unaware

28. Humans love the human component in everything; pure AI-generated content can only work through deception and scams or otherwise people wouldn’t consume it

29. To get good at AI you need to (1) be curious, (2) use the tools, and (3) have high openness to revisit your model of the world when AI does something weird

30. Your idea of AI will change over time—that’s okay