The five crazy days of OpenAI

The story that in recent days has involved OpenAI, the world’s largest artificial intelligence company, and Sam Altman, its co-founder and CEO, has been defined by many observers of the technology market as one of more daring. in recent years, and could have important consequences both for the company and for the future development of artificial intelligence (AI) technologies.

Within days, Sam Altman was fired from the company’s board amid enormous controversy and with little explanation, and the move generated enormous upheaval within the company. entire American technology sector: among other things, there was a general mutiny among employees and OpenAI was in danger of going bankrupt at the time of its greatest success. Then followed days of intense negotiations, which ultimately resulted in the return of Sam Altman as head of the company and the dismissal of almost the entire board of directors who had fired him.

What happened inside OpenAI is both a power struggle within one of the world’s richest and most promising companies and – according to many analysts who have covered it – a clash between two ways of conceiving the development of artificial intelligence and assessing the potential dangers of the technology: on the one hand Sam Altman, who has strongly insisted in recent years for rapid commercialization of the technologies developed by OpenAI, starting with ChatGPT ; on the other, the board of directors, which in all likelihood was pushing for more cautious development and greater attention to the hypothetical dangers that increasingly advanced artificial intelligence technologies could generate.

This shock arises from the history and very original structure of OpenAI, which is not a normal technological startup. It was founded in 2015 by Altman and other leading entrepreneurs and researchers (including Elon Musk and Peter Thiel) as a nonprofit that he had the task develop artificial intelligence technologies “in a way that can benefit humanity as a whole and does not require generating economic profits.” The need to control the development of artificial intelligence arises from the belief – very widespread among technological entrepreneurs – that the development of a “general artificial intelligence”, capable of performing practically any task almost on par with the he human being, can constitute an existential danger which must be evaluated with extreme caution.

In 2018, finding itself short of cash (among other things he said to himself that Elon Musk had tried to take control of the company, failed and then withdrew its funding), OpenAI split into two units: one remained non-profit and maintained its primary objective of developing non-profit artificial intelligence technologies harmful to humanity, while the other became an almost entirely for-profit company, which would operate like a normal business, attract investment, hire talented programmers and commercialize new products. The new for-profit unit (which, however, retained the name OpenAI) obtained a significant investment from Microsoft: initially one billion dollars, to which was later added 12. Under the new organization, the profits of the for-profit company would have been partly distributed. to investors but above all reinvested in research.

However, a relationship of dependence would have remained between the two units: the association’s board of directors, made up of four people, would have retained full control over the for-profit association, to ensure that the general objective of OpenAI – creating artificial intelligence that is not harmful to humanity – has been respected. The four people who make up the board of directors were, until a few days ago: Adam D’Angelo, co-founder of the question and answer site Quora, Ilya Sutskever, one of the co-founders of OpenAI and lead research team, Tasha McCauley, a technology entrepreneur, and Helen Toner, emerging technology security researcher.

OpenAI’s big growth moment came last year when, under pressure from CEO Sam Altman, the company’s for-profit unit took public ChatGPT, a highly advanced artificial intelligence software that became extremely popular in a very short time and brought OpenAI huge profits. visibility and significant investments. Largely thanks to ChatGPT and its subsequent developments, OpenAI is today one of the most valuable technology companies in the world, worth an estimated $86 billion.

But ChatGPT’s success has caused something of a division within the company. On the one hand, Sam Altman, supported by investors such as Microsoft, has pushed to accelerate and intensify the commercialization of new products that could guarantee new revenues and greater growth for OpenAI; on the other hand, other company executives, such as Ilya Sutskever, one of the board members, believed that the marketing of new products and research should be slowed down and subjected to more rigorous control.

Like, how he wrote L’Atlantic Within OpenAI, there is now an ongoing power struggle between the two ideological poles of society: a group born of Silicon Valley techno-optimism, energized by rapid commercialization [di nuovi prodotti]; the other convinced that artificial intelligence represents an existential risk for humanity and that it must be controlled with extreme caution.

In this state we arrive on Friday, November 17, when the board of directors of OpenAI (the non-profit) announced the immediate dismissal of Sam Altman. The board did not give specific reasons, but only issued a statement writing that Altman “had not always been honest” in his communications with the board.

The announcement surprised everyone: both OpenAI employees and Microsoft, which now has huge economic interests in the company. The reactions were immediate and almost all negative: Sam Altman is an extremely well-known figure in the American technological world, a charismatic and successful entrepreneur who is almost unanimously appreciated in the Silicon Valley knowledge system and to whom the success of ‘OpenAI. .

Within hours, some company executives, such as President Greg Brockman, resigned in protest. Microsoft also responded by harshly criticizing the board’s decision. The board of directors, however, remained on its position and appointed on Sunday evening, November 19, Emmett Shear, co-founder of the streaming site Twitch, critic of artificial intelligence and who had argued in the past that research was going slow motion and subject to much stricter controls. Meanwhile, Microsoft approached Altman and Brockman about creating a new unit responsible for artificial intelligence.

At that time, OpenAI employees staged an almost general mutiny against the board of directors: more than 700 of the 770 employees signed an open letter in which they threatened to follow Altman to Microsoft, among whom is even Ilya Sutskever, one of four board members, who said they regretted contributing to Altman’s firing.

At that point, new very intense negotiations began between Altman and the board of directors, which culminated on Tuesday evening: Altman returned to his position as CEO and the board of directors was almost completely removed and replaced by less intransigent figures, including the famous economist Larry Summers. . However, the board of directors obtained, according to various media outlets, that OpenAI open an internal investigation into Altman’s behavior. On Tuesday evening, a homecoming party for Altman was held at OpenAI headquarters in San Francisco until late at night.

Days later, it’s still unclear why the board suddenly decided to fire Altman. Reuters published an indiscretion that OpenAI was very close to major new breakthroughs in the field of artificial intelligence, which would greatly worry the board members. This indiscretion has, however, not been confirmed and, since the sources of the Reuters they are all internal to the company, it is also possible that this is a bit exaggerated.

Other analysts argued that Altman had engaged in behavior deemed unacceptable by the board of directors and that he had become too unscrupulous in the promises he made to investors and in the marketing of new products (which incidentally been numerous in recent months).

The most likely hypothesis is that Altman’s dismissal would be the culmination of the clash between the two positions present in the company in terms of artificial intelligence: the more cautious and the more aggressive.

Now that Altman and his more aggressive stance have actually won, it’s possible that OpenAI will change as well. The dispute with the board showed that the two-unit corporate structure, while designed to ensure the independence of research from profits, is actually still vulnerable to the will of large investors such as Microsoft , who could now play a more important role in “agency. The opinion many analysts is that with Altman’s return, OpenAI will increasingly abandon its non-profit goal and more decisively transform itself into a company like any other.

Source: ilpost


Leave a Reply

Your email address will not be published. Required fields are marked *