It may be a mistake to ask which occupations are most safe from being taken by AI. What AI (in its current form) is good at is not so much certain jobs, but a certain way of working. It's good at scutwork. So that's the thing to avoid. For example, is programming safe from AI? At the bottom end, definitely not. Those jobs are already disappearing. But at the same time the very best programmers (e.g. the ones who are good enough to start their own companies) are being paid exceptional amounts.

I was curious how well I have an essay loaded into my head when I'm working on it, so I tried reproducing the one I'm currently writing from memory. I reproduced it almost perfectly. But the differences suggest I don't have it memorized like lines in a play. Instead I seem to have it (or at least parts of it) stored as ideas rather than words. Mostly these come out the same, right down to the punctuation, but occasionally they turn into different words.

I just realized something most people are going to lose when (as they inevitably will) they start using AIs to write everything for them. They'll lose the knowledge of how writing is constructed. This is a very common kind of knowledge to be lost as technology progresses. Very few people now know how to weave cloth or turn pots or make baskets. But it's strange to think that writing will now enter this category.

The worst thing a pitch to professional investors can be is confusing. Professional investors are willing to give early stage startups the benefit of the doubt. They're ok with risk. But they can't invest in a startup whose pitch they can't even follow. It is therefore a big mistake for early stage startups to conceal the risks they face by being vague or evasive. You're hiding something that won't bother experts using something that will. Just admit the risks upfront, and explain why the EV is high nonetheless.

The classic software startup writes code to solve users' problems. If AI makes writing code more of a commodity, understanding users' problems will become the most important component of starting a startup. But it already is. What YC asks about in interviews is how well you understand users' needs, not your programming ability. I explained this years ago in this essay I wrote about how to ace your Y Combinator interview:

paulgraham.com/ace.html

What most people don't realize about Boom is that if they ship an airliner at all, every airline that flies internationally will have to buy it or be converted against their will into a discount airline, flying tourists subsonically. Ticket prices will be about the same as current business class prices on international flights. How can this be? Because the flights are so much shorter that you don't need lay-flat beds. You can use the seat pitch of domestic first class.

I just finished a new essay about the origins of wokeness. I'm going to get flak from both the far right and the far left about it, so to save all our time I'll respond in advance to what I know they'll say. People on the far right may, if they haven't been paying attention, claim that I'm just jumping on the bandwagon now that wokeness is already collapsing. In fact I've been writing about this topic since 2004, and I predicted this collapse a year ago:

https://x.com/paulg/status/1742333500621996409

AI startups seem to grow faster than previous startups. But this is just the latest stage in a consistent trend over the last 50 years. Each decade's startups grow faster than the previous decade's. Why does this happen? Startups don't just produce technology. They also consume technology. Better technology lets you do what you want faster. Startups want to grow. So better technology yields faster growth.

I have a file of notes about essays I might write in the future. Typically I'll have about a paragraph about each. I was curious how long the file was. 1792 lines! In practice this file is practically write-only. When I have an idea for an essay but don't have time to write it (e.g. because I'm in the middle of another) I make a note of it in the file. But I pretty much never go back and read them.

Want to start a startup doing eye-tracking? If so I'd be interested in funding it. A friend of mine has ALS and can only move his eyes. He has an eye-controlled keyboard, but it's not very good. Can you make him a better one? The reason this is worth doing as a startup is that I think eye-tracking may turn out to have broader applications than it currently does. It may even be a component of the next form factor in computing, the nature of which is still strangely unclear.

When you talk to more established founders, there's a common theme: late-stage investors pull some nasty shit. Once you get past series B you should stop optimizing for investors who can help you and switch to optimizing for ones who won't knife you if things go bad. One reason late stage investors behave so badly is that they're not worried about reputational damage the way seed and series A investors are. But YC may be able to make them start to worry. YC keeps a list of who's done what. Eventually that will affect their deal flow.

When I'm trying to help founders find new startup ideas, I usually start by trying to figure out what's unusual about them. What do they know or care about that few other people do? There's usually something, and it often leads to an idea. If this new idea was already inside them, why hadn't it bubbled to the surface? Two reasons: they didn't realize what was unusual about them, and they discounted it as a source of ideas.

Talked to a programmer today who said AI coding tools made him about 10x more productive. Though 10 seems like a round number, this was an attempt at a precise estimate. If that kind of increase in productivity were the norm, it would make starting a startup qualitatively different, because you could do with 8 programmers what would have taken 80, and a company with 8 programmers is much easier to fund and run than one with 80.

Observation suggests that people are switching to using ChatGPT to write things for them with almost indecent haste. Most people hate to write as much as they hate math. Way more than admit it. Within a year the median piece of writing could be by AI. I warn you now, this is going to have unfortunate consequences, just as switching to living in suburbia and driving everywhere did. When you lose the ability to write, you also lose some of your ability to think.

Oklahoma plans to kill Richard Glossip on May 18. The state's own attorney general has said that this would be a "grave injustice," but other people in Oklahoma are determined to go ahead and kill him. Sound crazy? Keep reading; it gets crazier. How did this happen? The usual mix of sloppy police work and corrupt prosecutors determined to win at any cost. The Innocence Project estimates that about 4% of people on death row are innocent. Richard Glossip is one of them.

Electric cars, fake meat, AI. All three have been around since I was a kid. They were never good enough, so we learned to ignore them. But all three have now crossed the line where you can't. What's next? As these examples suggest, "What's been around for a long time, but has never been good enough to take seriously?" is a good heuristic for predicting important new developments. When people have been trying to do something for a long time, there's usually a reason.

Amazon is the biggest used bookshop in the world, in the sense that they pack books so badly that any book you buy from them will be a used book by the time you get it. When I want to buy a book that's still in print, I begin by looking for a used copy in good condition on ABEbooks, since I know a reputable used bookshop will at least pack it well. (Yes, I know ABE is owned by Amazon.)

The amount of equity you'll get in a startup decreases exponentially as your employee number increases. If you want to get rich from a startup, you have to be there early, either as a founder or an early employee. But here's something few people get: startups need different kinds of work early on than they do later. So your chances of getting rich from a startup vary enormously depending on the type of work you do.

The underlying story, both here and with that BBC interview, is that there is now a whole generation who (a) have little respect for journalists after being treated dishonestly by them, and (b) no longer fear them, now that they've lost their monopoly on the news.

https://twitter.com/elonmusk/status/1646303780189900800

People who've grown up in this new world may think "Weren't things always this way?" No they were not. A generation ago, even the most powerful people used to try to stay on the right side of journalists. Very few dared to be openly contemptuous of them.

Jessica: What's new?

Me: OpenAI just released a new version that's much better. And people hadn't even stopped talking about the previous version yet.

Jessica: Are you scared about AI?

Me: A little. At the very least, it's going to change everything. One of my tests of the magnitude of a change is: How does it change what advice I'd give my kids? And when I try to answer that question, it's scary how little I can predict. The best I can do is tell them to surf this wave rather than be crushed by it.

When you're doing a deal with a large organization, find out if the people you're negotiating with actually have final say. Usually they don't, and that means the deal you've agreed upon can be, and often is, killed at the last minute by higher ups. Here's a simpler version that works just as well in practice: when making a deal with a large organization, no matter what they say, assume the deal will be cancelled at the last moment.