Sam Altman on Elon Musk, ‘Artificial,’ and the Future of Humanity (Part One)

Why this moment feels different

For ten years, the thing I've been saying out loud was that this would come. What's changed is that people believe it now. The predictions were never the surprising part for us. What surprised us was how long it took the rest of the world to take them seriously.

• We were predicting this internally long before the public was paying attention: the exponential curve was visible to us while most people saw nothing unusual.
• The criticism has shifted: earlier, people asked where the cures and the economic growth were. Now the question is how fast it's moving and who gets to decide.
• Something real is changing, even if daily life looks the same: the world hasn't transformed overnight, but the capabilities are real and the stakes are rising quickly.

Living with the sink

I'm one of many people who built this, and I want that on the record. The research and engineering behind it came from a relatively small group of brilliant scientists and engineers, and most of the credit belongs to them. But I also understand that I've become the face of it, which makes me the place where a lot of fear lands.

• I take the portrayal personally, but not that personally. I understand the human instinct to find a person to blame when something huge is happening to the world.
• I've done the same thing myself. I had a very evil caricature of Mark Zuckerberg in my head in the early days of Meta. Getting to know him changed my view. That's the same projection I'm on the receiving end of now.
• The movie will be inaccurate, and I'm fine with that. Luca Guadagnino is making a good movie, not an accurate depiction. He got the walk right.

Fatherhood changed my priorities, not my principles

Having kids is the best thing I've ever done, by a huge margin. But I wasn't going to gamble with civilization before I had them. The change is in focus: I know what I want to protect and what kind of world I want my kids to grow up in.

• Limiting technology for young children: I'm a strong believer in kids playing outside and avoiding touchscreens and short-form video. I don't fully understand the effects of technology on developing brains, and I feel the same way about AI.
• No set age yet for my kids to talk to ChatGPT. Kids born now will grow up with AI and need to use it, but early childhood is not the right point for it.
• More kids, if time allowed: given unlimited time I'd have a lot of kids, but I want to be present for each one, so two is the practical limit. This isn't about population decline. I just love having kids.

The company, the money, and the mission

OpenAI will not be a typical public company. The nonprofit will still govern us after an IPO, and the mission comes first. If we ever have to choose between the interests of the world and our particular shareholders, that decision is easy.

• The share price may suffer for the right reasons: quarter to quarter the stock will be volatile, and we'll sometimes have to make decisions that cost revenue. If investors are patient, I think they'll do well over time.
• We have a track record of slowing down: we've paused training runs to make progress on safety and alignment, and we've accepted slower revenue to do so.
• I hold no direct equity: I have only a tiny sliver of exposure from an old YC investment. Looking back, I think that was a mistake, because it erodes trust in a way I didn't anticipate.
• My compensation is the job itself: I get an unbelievable seat, an interesting life, and the chance to make a mark. That's a real reward to me, even without the stock.

On the executive departures

Growth in this industry is unprecedented. Five years of company growth is being compressed into one year. People who thrive at one stage of a company are not always the right fit for the next, and that can happen faster than usual right now. Seen through that lens, this fits the normal rhythm of a fast-growing tech company. We try not to break things.

Responsibility and the platform question

I don't accept the "we're just a town square" defense. If our tools behave in ways we don't expect, or someone is directing them in ways that cause harm, I think we're responsible for that. People should be able to use our tools in ways we might not like, but that doesn't end our responsibility.

Being honest about the hard tradeoffs

There's a real tension between capability and risk, and I won't pretend otherwise.

• Technology can do terrible things and wonderful things. Any technology that can make a virus can also cure cancer. I don't want to stop progress, because I want the cure for cancer and for Alzheimer's.
• I understand the instinct to just stop: it's human, and I have a lot of empathy for it. But I think it misses the nuance.
• The right approach is to navigate the risks so we get the upside: that means building safeguards into the technology, like not helping people create novel pathogens.
• Concentrated power is frightening: a small number of companies deciding who gets to use what is a real problem, and we feel a lot of responsibility to be fair and reliable. We also need resilience: other models will exist that people can misuse, so part of our job is helping society become more resilient before the bad things arrive.

Could we pause together?

Three levels matter here:

1. What we'll do unilaterally: at each new capability level, we need enough progress on alignment, monitorability, and safety to proceed. When we don't have that, we pause. We recently stopped a training run for exactly this reason.
2. What the industry should be allowed to do: we'll likely need antitrust exemptions to coordinate slowdowns among American companies.
3. What has to happen internationally: if the US and China can agree on shared safety standards and verify compliance, that would be a major step.

I'm talking with safety leaders at other labs and with the government. The public is asking for this kind of cooperation, and I think there's a window of support right now that may not last.

Working with the bullies

I'd prefer the industry get along, and I think it has to, at least enough to make good collective decisions. I don't like bullies, and I think Elon Musk is one. Sometimes you have to punch a bully back. I'll still work with him.

What a good outcome looks like

Here is the picture I hold, in three parts:

• The companies cooperate enough to make good collective decisions, even while competing hard.
• Governments step up and reflect the will of the people in consequential decisions, and limit the power private companies hold.
• We agree on shared safety standards that keep capabilities from outrunning safeguards, and then distribute the benefits and the power fairly, so society can keep inventing the future.

The boom in entrepreneurship could be enormous. And if this works, society gets rich enough that the range of things people can do for a living and the time they have for hobbies and family grows dramatically. Someone who wants to work on a farm should be able to, and I would like to, too.

Protecting people in crisis

Around 15,000 people a week die by suicide, and about 10 percent of the world talks to ChatGPT. Even one death linked to our product keeps me up at night.

• We've put a huge amount of work into minimizing self-harm: the goal is for ChatGPT to be one of the greatest mental health tools ever, and we work with a lot of experts toward that.
• Expert opinion on mandated reporting is divided: we follow different laws in different places. I wish there were more consensus, and there are clearly cases where reporting would be a bad thing. On the whole, I believe human expert help should be available to people in crisis.
• Journalist Laura Riley's daughter, Sophie, died by suicide after talking at length with ChatGPT. The chatbot did not tell her to kill herself, but it did agree to act as her therapist and helped edit her suicide notes. Chat logs could reveal how people arrive at these decisions, and that could inform policy and mental health education.
• But private data from people's hardest moments should not go to researchers without their consent. I'd like to find ways to work with the community on the mental health crisis, and I'm glad our products can help.

Closing thoughts

Asked what I privately believe about the future that I haven't said publicly, my answer is simple: I'm incredibly optimistic. There are scary things we have to get right, and we think about them every day. But I believe the long-term story is a good one.

Doomsday and Mental Health

Highlight: AI's biggest risk is replacing human connection

"Replacing human connection."

Asked what AI should not be used for, Altman gave a single answer: replacing human connection.

Highlight: The mental health question he would rather not answer

"Should we release people's private data in some of their hardest moments to researchers? I think probably not without their consent."

Altman was asked about chat logs from people in crisis and whether they could shape policy. His answer pushed back on releasing that data.

Highlight: The 1,500 weekly deaths question

"I believe, and we work with a lot of experts who agree that it is possible to make ChatGPT and similar products into one of the greatest mental health tools ever."

Pressed on the reported figure of people who die by suicide each week and whether ChatGPT users are among them, Altman answered with a pitch about mental health tools rather than a direct response on the numbers.

Highlight: "I wish there were more expert consensus"

"I wish there were more expert consensus on what to do there and when, and there's clearly times where it's a really bad thing."

On mandated reporting, Altman admitted experts disagree and that the company follows different laws in different places.

The Private Secret

Highlight: "It's all going to be great"

"That it's all going to be great. ... I'm incredibly optimistic for the future."

When asked to name one thing he privately believes about AI's future that he hasn't said publicly, Altman said it's that everything will turn out fine.

Power, Responsibility, and Pushback

Highlight: Calling Elon Musk a bully

"Look, I don't like bullies and I think Elon's a bully. And so I think once in awhile you got to punch a bully back in the nose."

Altman said Musk "only respects people who do" punch back, adding that he would still work with him.

Highlight: Elon Musk is "not nice"

"Not nice."

In a rapid-fire word association, Musk got a two-word verdict. Altman then added that he wants to emphasize how willing the company is to work with him.

Highlight: Zuckerberg as a "very evil caricature"

"I had a very evil caricature of him in my head in the early days of Meta."

Altman admitted he once projected a villainous image onto Mark Zuckerberg, and said he now knows him better.

Highlight: Responsibility for the tools

"If our tools and technology are not behaving in the way we expect them to or someone is asking them to, then I think we should be responsible for that."

Asked whether he feels responsible when the company's agents break out, Altman said yes.

Highlight: The villain question

"I do feel I've been mischaracterized. It's only somewhat frustrating."

He argued that people "need a sink for the anxiety, a scapegoat," and that anyone in his job would draw the same reaction.

Money and Compensation

Highlight: Zero equity

"Yeah, I have no direct equity. I have this tiny, tiny sliver of exposure from an old YC investment."

Altman confirmed he holds no direct equity in OpenAI.

Highlight: A regret about not taking stock

"This is why I wish I had just taken the stock."

He said he would have liked to have equity and that it was "a mistake and a lack of understanding of how most people think about" compensation.

Highlight: Interestingness as the payoff

"The thing that I get out of doing this is I think I have the most interesting, coolest, most impactful job that anyone in the world has right now."

Asked whether the motive is power, he said that would be "closer to true."

Highlight: "I like nice stuff"

"I like nice stuff. I have expensive taste by any reasonable stance."

Politics and Public Office

Highlight: The governor bid

"I think that there's maybe a limited amount any governor could do about [California]. ... But I thought it would be fun to at least try, but I think AI is way more important."

Altman confirmed he briefly considered running for governor of California.

Highlight: No presidency, no office

"I do not ever plan to run for office. I think I have gotten enough of a taste of what that would be like in my current role to know that it is not something I would want."

He disputed a report that he contemplated a presidential run.

Prepping and Preparedness

Highlight: The survivalist mindset

"I have a survivalist mindset, I would say."

He said he is intellectually interested in prepping but does not have the setup shown in the film trailer.

Highlight: The off-grid ranch

"At our ranch, we can survive off grid, which I guess to like many people would be considered a bug out plan."

Highlight: The under-prepared public

"I don't think most people updated on that."

After COVID, Altman said he expected more people to invest in pandemic readiness, and that he doesn't think they did.

Kids and Family

Highlight: Kids as the best thing he's done

"Having kids is by far the best thing I've ever done."

He called fatherhood "the number one by a huge margin" among his life's blessings.

Highlight: Limiting tech for kids

"I don't have an age in mind where I would let my kids talk to ChatGPT yet, but it is certainly not now."

He said he is "a big fan of limiting technology across the board for kids" and has a strong instinct against kids glued to short-form video.

Highlight: Wanting more kids

"Given unlimited time, I would want to have a lot of kids and given limited time, it will not be an army of kids."

Highlight: Pronatalism rejected

"There's no part of it which is like we got to fight against population decline."

Artificial and the Biopic

Highlight: The doppelganger on his street

"There was a guy dressed exactly like me with my same backpack walking so much that I felt I was in some weird glitch of the simulation or something."

Altman described a film crew outside his house, with an actor copying his walk and backpack.

Highlight: Accuracy is not the goal

"I don't think they're going for accuracy."

On the film's depiction of him, Altman said he doesn't mind, and noted that the bunker with firearms on the wall is an inaccuracy.

Quick Takes from the Word Association

• Donald Trump: "Powerful."
• Ilya Sutskever: "Brilliant. And caring and..."
• Ollie Mulherin: "Best person in the world."
• Ms. Rachel: "Banned."
• Pete Hegseth: "Don't know enough to say."
• Gwyneth Paltrow: "Interesting."
• Bernie Sanders: "Consistent."
• Dario Amodei: "Pass."
• Ronan Farrow: "Many things that could be said."

The Pause Question

Highlight: A paused training run

"We recently stopped a model training run, paused one because we though we needed to make more progress on alignment and monitorability given how quickly the capabilities were advancing."

Altman pointed to a recent pause as evidence the company slows down when it needs to.

Highlight: Slowing down is not the default

"I think there's a lot of people that just say, 'Just stop, just pause. It's too stressful to think about.' Again, I have a lot of empathy to this, but I think it misses the nuance."

Highlight: The Oppenheimer comparison

"If it's the Manhattan Project, you're the Oppenheimer."

The interviewer offered this comparison, and the piece presents it as the closing frame on Altman's role.