Sam Altman, the father of ChatGPT, superstar of the Davos forum

“This is the conference that everyone is waiting for,” proclaimed without modesty on Wednesday morning the star presenter of CNN Fareed Zakharia, regular at the World Economic Forum in Davos (WEF, in English). He stood on the stage of the plenary room where, for two days, Chinese Prime Minister Li Qiang, Ukrainian President Volodymyr Zelensky and even Emmanuel Macron have taken turns. Sam Altman, the founder of OpenAI, the start-up that invented ChatGPT, was indeed eagerly awaited by the approximately 1,200 spectators. To the point of stealing the spotlight from the other participants of the round table, all of them high caliber: three CEOs, Marc Benioff (Salesforce), Julie Sweet (Accenture) and Albert Bourla (Pfizer) and a minister, the chancellor of the exchequer British, Jeremy Hunt.

LISTEN TO OUR PODCAST

Appearing almost intimidated, the 38-year-old entrepreneur sat down, adjusting his pants. Asked about the high-profile psychodrama of his ouster and rapid reinstatement from OpenAI in November, Altman initially responded with an embarrassed smile and silence. “At a certain point, I think you just have to laugh about it, it’s so ridiculous,” he finally comments. We knew that our board of directors had become too small.” “What I learned the most was the strength of our team (...) which could have done without me,” he confided in front of a packed room. And slips that “with each new step towards powerful AI, the level of tension will increase, our degree of madness will rise by another ten points as the stakes are enormous”.

How does the creator of ChatGPT respond to concerns about generative artificial intelligence (AI), shared by Bill Gates, among others? “It’s a very powerful technology, we can’t say with certainty what will happen,” the person responds candidly. I understand the public’s nervousness about a company like ours.” “Not being careful, not understanding the seriousness of the issues would be a mistake,” he wants to reassure.

At his side, British Minister Jeremy Hunt points out that ChatGPT is still making mistakes that have damaged his ego: “The first time I went on ChatGPT, I asked: is Jeremy Hunt a good chancellor from the chessboard? ChatGPT told me that Jeremy Hunt was not chancellor. While emphasizing the progress of GPT3 and GPT4, Altman admits that the software still makes errors, and that users are much less tolerant of errors coming from a machine than from a human being.

To those, many, who worry that AI models are black boxes, Sam Altman responds that we manage to “x-ray the brain of an AI better than the human brain”. “What will be left for humans if AI calculates and analyzes better than them?” asks Fareed Zakharia. “Humans always need to know what other humans think and feel,” assures Altman. In my work, I am not an AI researcher, my role is above all to interact with other humans to make decisions.” “Take the example of chess,” he continues. When a machine beat Kasparov, we thought it was over.” But “no one watches a game of chess between two AIs” and “chess has never been more popular”.

As for the legitimate concern posed by the massive, and free, use of data to train language models, Sam Altman does not shy away from it. “We are in negotiations with the New York Times to pay for the use of their content,” he recalls. And in the future, “the models will rely on a smaller quantity of data but of better quality”.

If the plenary conference was open to all holders of the coveted white WEF badge, many Forum participants were looking for a smaller meeting with the boss of OpenAI. A “fad” that is causing a bit of cringe among Google and Meta executives, present in large numbers in Davos, and who are developing their own AI applications.

Some were able to approach Sam Altman on Tuesday afternoon during the “Bloomberg afternoon tea” organized by the American agency. On Wednesday, Microsoft, shareholder of OpenAI, invited around a hundred customers to attend an exchange between Satya Nadella, CEO of the giant founded by Bill Gates, and Sam Altman, in the Microsoft Café located opposite the convention center. A strategic location on the Promenade, the central street of Davos, where, for two kilometers, all the restaurants and shops (with the notable exception of the Rolex boutique and the branches of Credit Suisse and UBS) have been transformed into “ lounges” for the benefit of companies and countries.

Also readHow twenty French Tech stars want to ride the “Macron effect” in Davos

The Graal ? Obtain a bilateral meeting with Sam Altman. “I will do everything to get fifteen minutes face-to-face with him,” assured Tuesday Alex Combessie, CEO and co-founder of Giskard, a French start-up specializing in the certification of artificial intelligence algorithms against the risks of hallucinations. The ace. This turned out to be mission impossible, “even through French people working at OpenAI”, he regretted this Thursday after the conference. “He didn't make time in his busy schedule for Davos. This is only a postponement. We are going to San Francisco next month, we are going to try again,” reports the Frenchman.

Among the few to have had the privilege of a bilateral meeting: Antony Blinken, the US secretary of state, who saw him on Tuesday morning. As if the two Americans had to cross the Atlantic to meet. A bilateral, Christine Lagarde did not need it, as she knows Sam Altman well. At the end of Thursday's conference, the boss of the European Central Bank (ECB) simply took the stage to kiss him and exchange a few words. In Davos, there are few people who, sometimes out of snobbery, say they are not interested in a meeting. “I had lunch here with him last year,” confides a big boss. I didn’t find it that interesting.”