I spoke to Sam Altman of the launch fiasco of the GPT-5

Thursday, I dined with Sam AltmanSome other Openai leaders and a small group of journalists in San Francisco. Altman answered our questions for hours. No subject was prohibited, and everything, with the exception of what was said above the dessert, was in the file.

It is rare to have such an extensive and varied interview with a large CEO of technology on a meal. But there is nothing current in the situation in which Altman is. Chatgpt quickly became one of the most used influential products on Earth. Now Altman plots an aggressive expansion in consumer equipment, brain interfaces and social media. He is interested in buying Chrome if the US government forces Google to sell it. Oh, and he wants to raise billions of dollars to build data centers.

But first, he focuses on the response to last week’s deployment of GPT-5. About an hour before the start of dinner, Openai pushed an update to bring back the “heat” of 4o, its previous default model for chatgpt. It was Altman who called to quickly bring back 4o as an option to pay the subscribers after some have protested his disappearance on Reddit and X.

“I think we totally screwed up some things about deployment,” he said. “On the other hand, our API traffic has doubled in 48 hours and develops. We are out of the GPUs. Chatgpt has struck a new top of users every day. Many users really like the model switch. I think we learned a lesson on what it means upgrading a product for hundreds of millions of people in one day. ”

He set the percentage of chatgpt users who have unhealthy relationships with the product at “less than 1%”, but recognized that Openai employees “a lot” of meetings on the subject. “There are people who really felt like they had a relationship with Chatgpt, and the people we know and think. And then there are hundreds of millions of other people who do not have a parasocial relationship with Chatgpt, but who were very used to the fact that this has responded in a certain way, and would validate certain things, and would be favorable in certain ways. “

“You will certainly see some companies doing Japanese anime sex robots because they think they have identified something here that works,” he said in a not so subtle excavation in Grok. “You won’t see us doing it. We will continue to work hard to create a useful application, and we will try to allow users to use it as they wish, but not so much that people who have really fragile mental states are accidentally exploited.

Altman wants Chatgpt to feel as personal as possible but not necessarily to play with an ideology or a specific political vision. “I don’t think our products should be awake. I don’t think they should only be what is the opposite of that either. I think our product should have a fairly central center of the road, an intermediate position, and then you should be able to push it far enough. If you want: “I want you to be conservative”, it should think about you. “

Chatgpt has roughly quadrupled its user base in one year and now reaches more than 700 million people each week. “Very soon, billions of people a day will speak to Chatgpt,” said Altman. “We are the fifth website of the world right now. I think we are on the clear path to the third. ” (It means beating Instagram and Facebook.) “So it becomes more difficult. For Chatgpt to be bigger than Google, it’s really difficult.”

In order for its operation to continue to scale, Openai needs many more GPUs. This is one of Altman’s main priorities. “You should expect the OpenAi to spend billions of dollars for the construction of the data center in the not very distant future,” he said with confidence in the room.

“We have to make these horrible compromises at the moment,” he said. “We have better models, and we just can’t offer them because we don’t have the capacity. We have other types of new products and services that we would like to offer. ”

He also thinks that we are in an AI bubble. “When bubbles occur, intelligent people are excited about a core of truth,” he said. “If you look at most of the bubbles in history, such as the technological bubble, there was a real thing. The technology was really important. The Internet was really a big problem. People were overexcited. Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes. Is AI the most important thing that has been going on for a very long time? My opinion is also yes. “

He confirmed recent reports This OPENAI plans to finance a Brain-Complete interface startup to compete Elon Musk Neuralink. “I think the neural interfaces are nice ideas to explore. I would like to be able to think something and have the Chatppt to answer it. ”

Do Simo religions Ceivilling Ope To execute “applications”, will there be other autonomous applications in addition to chatgpt? “Yes, you should expect us from us.” He hinted at his Social media ambitions: “I am interested whether whether or not it is possible to create a much cooler type of social experience with AI.” He also said: “If Chrome will really sell, we should take a look at it.”

Although Altman has a lot of interest, it is not clear that running OPENAI in the long term is one. “I am not a naturally suitable person to be CEO of a public company,” he said at some point. “Can you imagine me during a call for results?”

I then asked if he would be CEO in a few years. “I mean, maybe AI is in three years. It’s long. “

Here are other things that Altman said:

Interesting career movements this week:

If you haven’t already done so, Don’t forget to subscribe to The peniswho understands Unlimited access to Command line And all our reports.

As always, I salute your comments. You can answer here or Ping me safely on the signal.