12 Pieces of Advice for Facing the AI Revolution

AI moved so fast this March that everyone caught a bad case of FOMO. Honestly, not catching it would have been impossible. In the past month alone, we got: Microsoft AI-ifying its entire product lineup at every tier (you can imagine this is hell for white-collar workers), Google's AI model, Baidu's AI model, and a parade of open-source AI models each claiming to be more power-efficient and easier to deploy than the last. And that's just the infrastructure layer. On the application layer, every image and graphics AI tool got a major update, especially the new releases from Adobe and Midjourney. The shift from "I don't quite feel it" to "I really feel it" is happening across the board — an AI era you can sense just by turning on your computer feels like it's finally arriving.

Fortunately, most of this is still just product launches for now, and plenty of people are mocking the idea that "the AI era hasn't arrived yet — the era of AI launch events has." But we all know AI is coming; what the big players unveiled at these events will show up on our computers before the year is out. We don't have to believe in the products the big vendors are announcing — but we should believe in their determination to prop up their stock prices!

A month later (yes, apparently that counts as a long time), here are twelve pieces of self-cultivation advice for facing the AI revolution, distilled from the anxious questions friends and family keep asking me.

Of course, who am I to lecture anyone on self-cultivation — but this is a synthesis of the AI column I write every single day, the mountain of material I read, and my experience selling AI products at work. I've been using this list to answer friends and family and the response has been pretty good, so I'm sharing it here. Each point could easily be expanded into its own essay.

Asking the right question matters — so does knowing how to break it down

1. Because the speed of generating answers is now measured in seconds, the ability to ask good questions matters more than ever. In the Google era, we asked questions to go find answers; now the answer might be wrong, but the speed and range of the response is a thousand times what it used to be. You no longer have to mentally translate the answer you found back into the question — the response already is a candidate answer.

2. Because asking questions matters, you need to be able to understand a problem and break it down. If you can't decompose a problem, it's hard to write a good prompt.

3. For the average person, prompt engineering is optional. From a software-sales perspective, prompt engineering is an intermediary technology whose shelf life as a discipline is uncertain. Also, AI itself is the best source of prompt-engineering technique — for now, ordinary people are still better off just learning to ask good questions.

4. Prompt-engineering "secrets" will mostly just raise your anxiety, without much practical payoff. The prompt templates floating around the market right now will likely be baked into the big vendors' products soon enough. And prompt tricks change dramatically with every new model version. If you really need an edge, I'd suggest just paying up — use your credit card to become a "whale" and run the strongest version of the software available.

5. Facing the AI giants doesn't mean you just lie down and take it. OpenAI isn't the only company on Earth that can build an AI model. This is an era to keep an open mind — don't laugh at anyone else's product, whatever it is. Two years ago, who would have guessed that the trendiest tech company today would be Microsoft? Also, if your job sits close to the technical foundation, you still need to keep building your own technical depth.

6. Data matters. Because general-purpose AI has already been consolidated by the giants, analyzing data has effectively become part of the AI commons available to everyone. That makes the data itself more precious. Collecting data will be important work going forward.

White-collar work will be affected — but there's no need for excessive pessimism

7. Debating today's AI limitations is pointless. Plenty of people love to ask AI a question, watch it fail, and then declare "see, AI can't do this." First, given the current pace of iteration, it will likely be able to do it very soon. Second, sure, AI has no hands and can't flip a burger — that's obvious and not worth discussing. The real problem is that this kind of behavior reeks of "back in my day" energy on social media — it reads as deeply uncool. We'd all rather be the cool uncle or aunt in the room...

8. Information literacy is the core skill of the future. Because different giants control different platforms, today's AI tools are already better at answering questions than at searching for them. So people with low information literacy will end up drowning in even more junk information. People who like fake news will find it a hundred times easier to get scammed going forward. You need to build a rigorous habit of screening out misinformation — and help your family build that same literacy.

9. Basic literacy and numeracy are still required. Without a baseline education, you can't even formulate a good question.

10. Existing white-collar jobs will be seriously affected. This point deserves its own column. But I think Bill Gates is being overly optimistic here — existing white-collar work will be hit hard by AI. AI can already do most of today's white-collar tasks better than humans, and the remaining gaps will close as the technology keeps iterating.

11. New white-collar jobs are nothing to be optimistic about. Following on from the last point — we haven't yet spotted any new category of white-collar work that can immediately absorb the people displaced by this wave. Some people say "prompt engineer." I (and plenty of people smarter than me) think that's just an intermediary technology, unlikely to generate anywhere near enough demand.

12. You must embrace change. Even if we do nothing at all, AI will keep shipping new versions regardless. Going forward, every decision has to be made on the premise that "the AI era has already arrived and there's no driving it away." You need that mindset — the only option is to embrace the change.

That's my preliminary conclusion — offered humbly, for what it's worth.