GPT-4o

There are two things from our announcement today I wanted to highlight.

First, a key part of our mission is to put very capable AI tools in the hands of people for free (or at a great price). I am very proud that we’ve made the best model in the world available for free in ChatGPT, without ads or anything like that. 

Our initial conception when we started OpenAI was that we’d create AI and use it to create all sorts of benefits for the world. Instead, it now looks like we’ll create AI and then other people will use it to create all sorts of amazing things that we all benefit from. 

We are a business and will find plenty of things to charge for, and that will help us provide free, outstanding AI service to (hopefully) billions of people. 

Second, the new voice (and video) mode is the best computer interface I’ve ever used. It feels like AI from the movies; and it’s still a bit surprising to me that it’s real. Getting to human-level response times and expressiveness turns out to be a big change.

The original ChatGPT showed a hint of what was possible with language interfaces; this new thing feels viscerally different. It is fast, smart, fun, natural, and helpful.

Talking to a computer has never felt really natural for me; now it does. As we add (optional) personalization, access to your information, the ability to take actions on your behalf, and more, I can really see an exciting future where we are able to use computers to do much more than ever before.

Finally, huge thanks to the team that poured so much work into making this happen!

What I Wish Someone Had Told Me

  1. Optimism, obsession, self-belief, raw horsepower and personal connections are how things get started.
  2. Cohesive teams, the right combination of calmness and urgency, and unreasonable commitment are how things get finished. Long-term orientation is in short supply; try not to worry about what people think in the short term, which will get easier over time.
  3. It is easier for a team to do a hard thing that really matters than to do an easy thing that doesn’t really matter; audacious ideas motivate people.
  4. Incentives are superpowers; set them carefully.
  5. Concentrate your resources on a small number of high-conviction bets; this is easy to say but evidently hard to do. You can delete more stuff than you think.
  6. Communicate clearly and concisely.
  7. Fight bullshit and bureaucracy every time you see it and get other people to fight it too. Do not let the org chart get in the way of people working productively together.
  8. Outcomes are what count; don’t let good process excuse bad results.
  9. Spend more time recruiting. Take risks on high-potential people with a fast rate of improvement. Look for evidence of getting stuff done in addition to intelligence.
  10. Superstars are even more valuable than they seem, but you have to evaluate people on their net impact on the performance of the organization.
  11. Fast iteration can make up for a lot; it’s usually ok to be wrong if you iterate quickly. Plans should be measured in decades, execution should be measured in weeks.
  12. Don’t fight the business equivalent of the laws of physics.
  13. Inspiration is perishable and life goes by fast. Inaction is a particularly insidious type of risk.
  14. Scale often has surprising emergent properties.
  15. Compounding exponentials are magic. In particular, you really want to build a business that gets a compounding advantage with scale.
  16. Get back up and keep going.
  17. Working with great people is one of the best parts of life.

Helion Needs You

Helion has been progressing even faster than I expected and is on pace in 2024 to 1) demonstrate Q > 1 fusion and 2) resolve all questions needed to design a mass-producible fusion generator.

The goals of the company are quite ambitious—clean, continuous energy for 1 cent per kilowatt-hour, and the ability to manufacture enough power plants to satisfy the current electrical demand of earth in a ten year period.

If both things happen, it will transform the world. Abundant, clean, and radically inexpensive energy will elevate the quality of life for all of us—think about how much the cost of energy factors into what we do and use. Also, electricity at this price will allow us to do things like efficiently capture carbon (so although we’ll still rely on gasoline for awhile, it’ll be ok).

Although Helion’s scientific progress of the past 8 years is phenomenal and necessary, it is not sufficient to rapidly get to this new energy economy. Helion now needs to figure out how to engineer machines that don’t break, how to build a factory and supply chain capable of manufacturing a machine every day, how to work with power grids and governments around the world, and more.

The biggest input to the degree and speed of success at the company is now the talent of the people who join the team. Here are a few of the most critical jobs, but please don’t let the lack of a perfect fit deter you from applying.

Electrical Engineer, Low Voltage: https://boards.greenhouse.io/helionenergy/jobs/4044506005
Electrical Engineer, Pulsed Power: https://boards.greenhouse.io/helionenergy/jobs/4044510005
Mechanical Engineer, Generator Systems: https://boards.greenhouse.io/helionenergy/jobs/4044522005
Manager of Mechanical Engineering: https://boards.greenhouse.io/helionenergy/jobs/4044521005

DALL•E 2

Today we did a research launch of DALL•E 2, a new AI tool that can create and edit images from natural language instructions. 

Most importantly, we hope people love the tool and find it useful. For me, it’s the most delightful thing to play with we’ve created so far. I find it to be creativity-enhancing, helpful for many different situations, and fun in a way I haven’t felt from technology in a while.

But I also think it’s noteworthy for a few reasons:

1) This is another example of what I think is going to be a new computer interface trend: you say what you want in natural language or with contextual clues, and the computer does it. We offer this for code and now image generation; both of these will get a lot better. But the same trend will happen in new ways until eventually it works for complex tasks—we can imagine an “AI office worker” that takes requests in natural language like a human does.

2) It sure does seem to “understand” concepts at many levels and how they relate to each other in sophisticated ways.

3) Copilot is a tool that helps coders be more productive, but still is very far from being able to create a full program. DALL•E 2 is a tool that will help artists and illustrators be more creative, but it can also create a “complete work”. This may be an early example of the impact AI on labor markets. Although I firmly believe AI will create lots of new jobs, and make many existing jobs much better by doing the boring bits well, I think it’s important to be honest that it’s increasingly going to make some jobs not very relevant (like technology frequently does).

4) It’s a reminder that predictions about AI are very difficult to make. A decade ago, the conventional wisdom was that AI would first impact physical labor, and then cognitive labor, and then maybe someday it could do creative work. It now looks like it’s going to go in the opposite order.

5) It’s an example of a world in which good ideas are the limit for what we can do, not specific skills.

6) Although the upsides are great, the model is powerful enough that it's easy to imagine the downsides.

Hopefully this summer, we’ll do a product launch and people will be able to use it for all sorts of things. We wanted to start with a research launch to figure out how to minimize the downsides in collaboration with a larger group of researchers and artists, and to give people some time to adapt to the change—in general, we are believers in incremental deployment strategies. (Obviously the world already has Photoshop and we already know that images can be manipulated, for good and bad.)

 (A robot hand drawing, by DALL•E)


Helion

I’m delighted to be investing more in Helion. Helion is by far the most promising approach to fusion I’ve seen.

David and Chris are two of the most impressive founders and builders (in the sense of building fusion machines, in addition to building companies!) I have ever met, and they have done something remarkable. When I first invested in them back in 2014, I was struck by the thoughtfulness of their plans about the scientific approach, the system design, cost optimizations, and the fuel cycle.

And now, with a tiny fraction of the money spent on other fusion efforts but the culture of a startup, they and their team have built a generator that produces electricity. Helion has a clear path to net electricity by 2024, and has a long-term goal of delivering electricity for 1 cent per kilowatt-hour. (!)

If this all works as we hope, we may have a path out of the climate crisis. Even though there are a lot of emissions that don’t come from electrical generation, we’d be able to use abundant energy to capture carbon and other greenhouses gases.

And if we have much cheaper energy than ever before, we can do things that are difficult to imagine today. The cost of energy is one of the fundamental inputs in the costs of so much else; dramatically cheaper energy will lead to dramatically better quality of life for many people.

The Strength of Being Misunderstood

A founder recently asked me how to stop caring what other people think. I didn’t have an answer, and after reflecting on it more, I think it's the wrong question.

Almost everyone cares what someone thinks (though caring what everyone thinks is definitely a mistake), and it's probably important. Caring too much makes you a sheep. But you need to be at least a little in tune with others to do something useful for them.

It seems like there are two degrees of freedom: you can choose the people whose opinions you care about (and on what subjects), and you can choose the timescale you care about them on. Most people figure out the former [1] but the latter doesn’t seem to get much attention.

The most impressive people I know care a lot about what people think, even people whose opinions they really shouldn’t value (a surprising numbers of them do something like keeping a folder of screenshots of tweets from haters). But what makes them unusual is that they generally care about other people’s opinions on a very long time horizon—as long as the history books get it right, they take some pride in letting the newspapers get it wrong. 

You should trade being short-term low-status for being long-term high-status, which most people seem unwilling to do. A common way this happens is by eventually being right about an important but deeply non-consensus bet. But there are lots of other ways–the key observation is that as long as you are right, being misunderstood by most people is a strength not a weakness. You and a small group of rebels get the space to solve an important problem that might otherwise not get solved.


 

[1] In the memorable words of Coco Chanel, “I don’t care what you think about me. I don’t think about you at all.”

PG and Jessica

A lot of people want to replicate YC in some other industry or some other place or with some other strategy. In general, people seem to assume that: 1) although there was some degree of mystery or luck about how YC got going, it can’t be that hard, and 2) if you can get it off the ground, the network effects are self-sustaining.

More YC-like things are good for the world; I generally try to be helpful. But almost none of them work. People are right about the self-sustaining part, but they can’t figure out how to get something going.

The entire secret to YC getting going was PG and Jessica—there was no other magic trick. A few times a year, I end up in a conversation at a party where someone tells a story about how much PG changed their life—people speak with more gratitude than they do towards pretty much anyone else. Then everyone else agrees, YC founders and otherwise (non-YC founders might talk about an impactful essay or getting hired at a YC company). Jessica still sadly doesn’t get nearly the same degree of public credit, but the people who were around the early days of YC know the real story.

What did they do? They took bets on unknown people and believed in them more than anyone had before. They set strong norms and fought back hard against bad behavior towards YC founders. They trusted their own convictions, were willing to do things their way, and were willing to be disliked by the existing power structures. They focused on the most important things, they worked hard, and they spent a huge amount of time 1:1 with people. They understood the value of community and long-term orientation. When YC was very small, it felt like a family.

Perhaps most importantly, they built an ecosystem (thanks to Joe Gebbia for pointing this out). This is easy to talk about but hard to do, because it requires not being greedy. YC has left a lot of money on the table; other people have made more money from the ecosystem than YC has itself. This has cemented YC’s place—the benefits to the partners, alumni, current batch founders, Hacker News readers, Demo Day investors, and everyone else around YC is a huge part of what makes it work.

I am not sure if any of this is particularly useful advice—none of it sounds that hard, and yet in the 15 years since, it hasn’t been close to replicated.

But it seems worth trying. I am pretty sure no one has had a bigger total impact on the careers of people in the startup industry over that time period than the two of them.

Researchers and Founders

I spent many years working with founders and now I work with researchers.

Although there are always individual exceptions, on average it’s surprising to me how different the best people in these groups are (including in some qualities that I had assumed were present in great people everywhere, like very high levels of self-belief).

So I’ve been thinking about the ways they’re the same, because maybe there is something to learn about qualities of really effective people in general.

The best people in both groups spend a lot of time reflecting on some version of the Hamming question—"what are the most important problems in your field, and why aren’t you working on them?” In general, no one reflects on this question enough, but the best people do it the most, and have the best ‘problem taste’, which is some combination of learning to think independently, reason about the future, and identify attack vectors. (This from John Schulman is worth reading: http://joschu.net/blog/opinionated-guide-ml-research.html).

They have a laser focus on the next step in front of them combined with long-term vision. Most people only have one or the other.

They are extremely persistent and willing to work hard. As far as I can tell, there is no high-probability way to be very successful without this, and you should be suspicious of people who tell you otherwise unless you’d be happy having their career (and be especially suspicious if they worked hard themselves).

They have a bias towards action and trying things, and they’re clear-eyed and honest about what is working and what isn’t (importantly, this goes both ways—I’m amazed by how many people will see something working and then not pursue it). 

They are creative idea-generators—a lot of the ideas may be terrible, but there is never a shortage.

They really value autonomy and have a hard time with rules that they don’t think make sense. They are definitely not lemmings.

Their motivations are often more complex than they seem—specifically, they are frequently very driven by genuine curiosity.

Project Covalence

Almost every company and non-profit working on COVID-19 that I offered to help asked for support with clinical trials—for companies focusing on developing novel drugs, vaccines, and diagnostics, rapidly spinning up trials is one of their biggest bottlenecks. 

Science remains the only way out of the COVID-19 crisis. Dramatically improving clinical trials, which are usually time-consuming and cost tens to hundreds of millions of dollars, is one of the highest-leverage ways to get out of it faster.  

The goal of this project, in collaboration with TrialSpark and Dr. Mark Fishman, is to offer much better clinical trial support to COVID-19 projects than anything that currently exists.

Project Covalence’s platform, powered by TrialSpark, is uniquely optimized to support COVID-19 trials, which are ideally run in community settings or at the patient’s home to reduce the burden placed on hospitals and health systems. Project Covalence is well-positioned to tackle the operational and logistical challenges involved in launching such trials, and supports trial execution, 21 CFR Part 11 compliant remote data collection, telemedicine, biostatistics, sample kits for at-home specimen collection, and protocol writing. 

Researchers across academia and industry can leverage this shared infrastructure to rapidly launch their clinical trials. To facilitate coordination between studies, we will also be creating master protocols for platform studies to enable shared control arms and adaptive trial designs.

If you’re interested in getting involved or have a trial that needs support, please get in touch at ProjectCovalence@trialspark.com or visit www.projectcovalence.com.

Idea Generation

The most common question prospective startup founders ask is how to get ideas for startups. The second most common question is if you have any ideas for their startup.

But giving founders an idea almost always doesn’t work. Having ideas is among the most important qualities for a startup founder to have—you will need to generate lots of new ideas in the course of running a startup.

YC once tried an experiment of funding seemingly good founders with no ideas. I think every company in this no-idea track failed. It turns out that good founders have lots of ideas about everything, so if you want to be a founder and can’t get an idea for a company, you should probably work on getting good at idea generation first.

How do you do that?

It’s important to be in the right kind of environment, and around the right kind of people. You want to be around people who have a good feel for the future, will entertain improbable plans, are optimistic, are smart in a creative way, and have a very high idea flux. These sorts of people tend to think without the constraints most people have, not have a lot of filters, and not care too much what other people think. 

The best ideas are fragile; most people don’t even start talking about them at all because they sound silly. Perhaps most of all, you want to be around people who don’t make you feel stupid for mentioning a bad idea, and who certainly never feel stupid for doing so themselves.

Stay away from people who are world-weary and belittle your ambitions. Unfortunately, this is most of the world. But they hold on to the past, and you want to live in the future.

You want to be able to project yourself 20 years into the future, and then think backwards from there. Trust yourself—20 years is a long time; it’s ok if your ideas about it seem pretty radical. 

Another way to do this is to think about the most important tectonic shifts happening right now. How is the world changing in fundamental ways? Can you identify a leading edge of change and an opportunity that it unlocks? The mobile phone explosion from 2008-2012 is the most recent significant example of this—we are overdue for another!

In such a tectonic shift, the world changes so fast that the big incumbents usually get beaten by fast-moving and focused startups. (By the way, it’s useful to get good at differentiating between real trends and fake trends. A key differentiator is if the new platform is used a lot by a small number of people, or used a little by a lot of people.)

Any time you can think of something that is possible this year and wasn’t possible last year, you should pay attention. You may have the seed of a great startup idea. This is especially true if next year will be too late.

When you can say “I am sure this is going to happen, I’m just not sure if we’ll be the ones to do it”, that’s a good sign. Uber was like this for me—after the first time I used it, it was clear we weren’t going to be calling cabs for that much longer, but I wasn’t sure that Uber was going to win the space.

A good question to ask yourself early in the process of thinking about an idea is “could this be huge if it worked?” There are many good ideas in the world, but few of them have the inherent advantages that can make a startup massively successful. Most businesses don’t generate a valuable accumulating advantage as they scale. Think early about why an idea might have that property. It’s obvious for Facebook or Airbnb, but it often exists in more subtle ways.

It’s also important to think about what you’re well-suited for. This is hard to do with pure introspection; ideally you can ask a mentor or some people you’ve worked with what you’re particularly good at. I’ve come to believe that founder/company fit is as important as product/market fit.

Finally, a good test for an idea is if you can articulate why most people think it’s a bad idea, but you understand what makes it good.


This is from my notes for a talk I gave at a YC event in China in 2018. Thanks to Eric Migicovsky for encouraging me to post it!

I wrote it when I thought mostly about startups; now I think mostly about AI development. I am struck by how much of it applies, particularly paragraphs 5-9.