The creator of ChatGPT pleads for “a fair balance” between regulation and innovation

MISE À DAY

Will or not leave the European Union (EU)? Passing through Paris on Friday, the boss of OpenAI and creator of ChatGPT, the American Sam Altman, assured that he did not intend to stop operating in Europe, but he demanded from the EU a “fair balance” between protection and innovation. 

During an exchange on artificial intelligence and the future of media at Station F, a nursery for young companies, the 38-year-old leader explained that he discussed how to find “the right balance between protection and positive impact” of this technology with President Emmanuel Macron on Tuesday.

“We plan to comply” with the future European regulation on AI (IA Act) and “we really like Europe”, but “we want to make sure we are technically capable of it,” he said.

“An authorization regime for the general framework and security standards are very relevant,” he continued. “But to say, when you don't know how generative AI works, 'you have to meet this guarantee 100% of the time', honestly, we don't know how to do that.”

On Wednesday, during the London stage of his international tour to reassure about fears linked to AI (disinformation, destruction of jobs, looting of works, etc.), Sam Altman had threatened to leave the EU if the regulation was becoming too restrictive there, triggering the anger of European Commissioner Thierry Breton, who cried out for “blackmail”.

Friday morning, the creator of ChatGPT sought appeasement. “Very productive week of conversations in Europe on how best to regulate AI! We are delighted to continue operating here and of course have no intention of leaving,” he tweeted.

In the afternoon, he praised the reception of France, “a very interesting case, much more advanced in this technology and in its adoption than other countries”, he launched, in front of an audience including many representatives of young companies.

However, he refrained from saying whether he intended to set up a headquarters in the country or elsewhere in Europe: “We will open headquarters around the world, but very slowly, we're still a small company.”

“How not to be scared?”

“In the audience, Louis Dreyfus, president of the executive board of Le Monde, questioned him, to applause. “My business model is to pay talented journalists to produce content and have people pay for it. I talked to other publishers, who see AI as producing content without any human interaction and for free. Can you tell me how not to be scared off by this and what my business model will be in the future?”

“You are one of the biggest newspapers in the world and I don't think you're going to be duplicated by AI anytime soon,” replied Sam Altman. “But I bet your journalists can use AI in their creative process, do better investigations, come up with better ideas.”

“There's something deep about human taste and about humans who know what other humans want and ask questions. And people want to know which journalist wrote and to have a connection with the person who created, even if there could be great texts written by AI,” he added.

Then questioned by a small group of journalists, including AFP, the American leader confided in his feelings about the dazzling success of OpenAI, whose AIs (ChatGPT for text and Dall-E for images) have opened to the general public at the end of 2022.

“In a few years, I will feel like I have lived a very special moment… but it is also very exhausting and I hope the life will calm down,” he said.

He also defended his company, which is often criticized for not publishing the sources of its training corpus, which may consist of works subject to copyright or contain illegal or hateful content.

Sam Altman argued that the critics mostly wanted to check whether the models themselves were racist. “What matters is the result of the racial bias test,” he said, refuting the idea that he should publish his sources and adding that his latest model, GPT-4, was “surprisingly not biased”.

Confronted with recent statements by billionaire Elon Musk, who considers that OpenAI – which he had financed until 2018 – has betrayed its promise not to seek profit and is now managed from made by Microsoft, the creator of chatGPT said he “disagrees with almost everything, but I'll try to avoid getting into a pitched battle”.

Sam Altman has finally confirmed that he didn't own a share of the company himself, unlike other big tech executives.