Sam Altman on how to start a startup
Altman argues you should only start a startup when a specific problem compels you, ideally with an idea that sounds bad but is good in a small, fast-growing market. Success then comes down to making a small group of users love the product enough to spread it themselves, and refusing to quit before the work has had time to compound. By 2025-26 he adds that AI makes this an unusually good moment to start: costs and cycle times are collapsing, so founders should bet on continued exponential model progress and build differently from the received wisdom.
11 points · 5 appearances · 2014–2026
How the view evolved
Make a few users love it → Spontaneous word-of-mouth is 80% of success
Same underlying bar, but the framing hardens between 2014 and 2018: in 2014 intense love from a small user group is one prerequisite — the thing to expand from — while by 2018 he restates it as spontaneous word-of-mouth and quantifies it as '80% of the work,' making it the near-definition of startup success rather than step one. A sharpening of emphasis, not a reversal.
Up next
2023–2026 · changed view
“I love that you asked them together, because I think they're super related. I think the economic transformation will drive much of the political transformation here, not the other way around. My working model for the last five years has been that the two dominant changes will be that the cost of intelligence and the cost of energy are going, over the next couple of decades, to dramatically, dramatically fall from where they are today. And the impact of that, and you're already seeing it with the way you now have like, you know, programming ability beyond what you had as an individual before, is society gets much, much richer, much wealthier, in ways that are probably hard to imagine. I think every time that's happened before, it has been that economic impact has had positive political impact as well. And I think it does go the other way too, like the socio-political values of the Enlightenment enabled the long-running technological revolution and scientific discovery process we've had for the past centuries. But I think we're just going to see more. I'm sure the shape will change, but I think it's just long and beautiful exponential curve.”
2023–2026 · changed view
“So I want to be very clear. I do not think we have yet discovered a way to align a super powerful system. We have something that works for our current scale called RLHF, and we can talk a lot about the benefits of that and the utility it provides. It's not just an alignment, maybe it's not even mostly an alignment capability. It helps make a better system, a more usable system. And this is actually something that I don't think people outside the field understand enough. It's easy to talk about alignment and capability as orthogonal vectors. They're very close. Better alignment techniques lead to better capabilities, and vice versa. There's cases that are different, and they're important cases, but on the whole, I think things that you could say like RLHF or interpretability that sound like alignment issues also help you make much more capable models, and the division is just much fuzzier than people think. And so in some sense, the work we do to make GPT-4 safer and more aligned looks very similar to all the other work we do of solving the research and engineering problems associated with creating useful and powerful models.”
2023–2026 · changed view
“People talk about like how many jobs is AI gonna do in five years and the framework that people have is what percentage of current jobs are just gonna be totally replaced by some AI doing the job? The way I think about it is not what percent of jobs AI will do, but what percent of tasks will AI do and over what time horizon. So if you think of all of the like-five second tasks in the economy, five-minute tasks, the five-hour tasks, maybe even the five-day tasks, how many of those can AI do? And I think that's a way more interesting, impactful, important question than how many jobs AI can do because it is a tool that will work at increasing levels of sophistication and over longer and longer time horizons for more and more tasks and let people operate at a higher level of abstraction. So maybe people are way more efficient at the job they do. And at some point, that's not just a quantitative change, but it's a qualitative one too about the kinds of problems you can keep in your head. I think that for videos on YouTube, it'll be the same. Many videos, maybe most of them, will use AI tools in the production, but they'll still be fundamentally driven by a person thinking about it, putting it together, doing parts of it, sort of directing it and running it.”
2023–2026 · changed view
“First of all, you can't ask it anything. That's very nice of you to say, but there's a lot of things that it's still embarrassingly bad at. But even if we fixed those, which hopefully we will, it doesn't continuously learn and improve. It can't go get better at something that it's currently weak at. It can't go discover new science and update its understanding and do that. And it also kind of can't — even if we lower the bar — it can't just sort of do any knowledge work you could do in front of a computer. Even without the sort of ability to get better at something it doesn't know yet, I might accept that as a definition of AGI. But the current systems — you can't say, like, hey, go do this task for my job, and it goes off and clicks around the internet and calls someone and looks at your files and does it. And without that, it feels definitely short of it.”