Caught up in life and death in a matter of hours?Why was Ultraman, the soul of OpenAI, deposed?
Xu Xu
The reason for quitting is like the reason for a lover's breakup, and the high-sounding words are not the real reason, and the truth is often more realistic and cruel.
OpenAI is the strongest AI (artificial intelligence) research company on the surface, and its founder, Sam Altman, is often translated by the media as Sam Altman, representing a new human with superpowers and potential comparable to that of "alien" Elon Musk. As a result, in the middle of the night on November 18, he was suddenly "expelled" by the OpenAI board of directors.
The reason given by the Board of Directors is that "prior to Altman's departure, the Board of Directors conducted a careful review and concluded that he had not always been open in his communications with the Board of Directors, which hampered the Board's ability to perform its duties." The board no longer has confidence in his ability to continue to lead Open AI. ”
This sentence is still to be said, and it contains too much information and subtext.
What did Ultraman not always be honest about? What was his original intention? Was he shaken as the founder? What was the disagreement between the board of directors and him?
At the same time, Greg Brockman, Altman's close associate, co-founder and chairman of the board, is also out of OpenAI. This means that the founding team that rallied around Ultraman and Brockman could be in turmoil.
Wang Bing, chairman of Dingtian Asset Management Co., Ltd., said on Xiaohongshu: "As I imagined, the internal strife of startups is deteriorating very quickly. Last week was also the OpenAI Dev day of the Spring Festival Gala of Science and Technology. Now that Sam has been fired, Greg has resigned as founder and president after receiving the news. Greg was formerly the CTO of stripe. The OpenAI GTM team is all the best players before Stripe. OpenAI is now the strongest from the surface of the large language model, and within three hours it is already in jeopardy. This is entrepreneurship, even if you are valued at hundreds of millions, survival is instantaneous. ”
At present, OpenAI is more like a palace fight drama has taken place, and Ultraman has been expelled. This aspect is related to the company's shareholding structure, Ultraman does not have absolute power over the company, which led to the founder being ousted.
On the other hand, the essence of this conflict comes from the dispute over how to balance OpenAI's mission, vision, values and commercial profits, AI is a huge controversial business involving the "destruction and survival of all mankind", the company's commercial profits and mission, vision and values are different, even contradictory, where is the boundary between OpenAI as a non-profit organization and a commercial company?
The inability to balance these should be the main reason why Ultraman is out as the soul of the company.
OpenAI创始人Sam Altman。
Conflict between commercial profits and non-profit organizations
OpenAI's original intention and mission was to use AI to save the world, not destroy it. It is this vision and values that attracts many scientists to join.
Founded in Silicon Valley in 2015 as a non-profit research lab, OpenAI's goal is to "advance AI in a way that benefits society as a whole, without being constrained by the need for financial returns."
But then, OpenAI faced two big challenges, one was external: a non-profit organization, how to solve the problem of high budgets. Training a large model is an extremely money-burning game.
OpenAI started out with donations. For example, OpenAI finally received a donation of $130.5 million, of which Elon Musk, CEO of SpaceX, as a co-chair, donated $100 million. Donation money is a drop in the bucket for the huge cost of research.
Another challenge comes from within: how to balance the original aspirations of AI, the most valuable technological capability of the 21st century, with the long-term struggle to bring huge commercial profits.
At the beginning of 2019, Musk, the biggest financier of OpenAI, broke up because of a disagreement with the founding team. Altman officially took over OpenAI and started a major transformation of OpenAI from a non-profit organization to a for-profit enterprise. This can also show Ultraman's ambition in business.
This has also drawn huge criticism, with Ultraman's actions undermining the original moral commitment to the development of AI.
This also meant that founders Altman and Brockman needed to balance the pursuit of commercial profit with an ethical commitment to the values of the original mission, vision, and values.
It can't be said that Ultraman didn't try to balance the trade-offs between the two. For example, OpenAI has set up a "capped profit" framework that imposes restrictions on first-round investors, authorizing them to only reap the maximum return of 100 times their original investment.
But it is clear that the internal differences have not been balanced.
According to an article from Tencent News's "Frontline", several board members represented by OpenAI's chief scientist IIya Sutskever believe that Ultraman's push is too much and too fast, and it is not the right path to achieve general artificial intelligence. Chief Scientist IIya Sutskever had serious disagreements with Altman and Brockman, and eventually he managed to convince the support of three other board members (a total of six members of the board of directors) to obtain a decisive majority of votes to expel Altman from the company.
Thus creating this sensational event of the founder's outing.
It has become a controversy and choice in the entire AI field
However, after all, OpenAI's research results involve industrial transformation, technology monopoly, social responsibility, etc., which will have a profound impact on the economy and society in the coming decades, accompanied by huge commercial returns.
With the birth of ChatGPT, this conflict has actually broken through OpenAI and risen to a controversy and choice in the entire AI field.
In 2020, OpenAI developed ChatGPT, which broke the ground, and the IT industry seemed to realize that this is a "steam engine"-like era of technological change. The camp of AI development is also being torn apart. On the one hand, Musk feels that artificial intelligence is summoning demons and rallying a group of opponents to resist the accelerated development of AI.
On the other hand, the world, including China, has set off a "great leap forward" in the development of large AI models, and since April, there has been a press conference of technology companies almost every day to announce the birth of large models or announce the birth of large models.
Is it possible for OpenAI, the company that Ultraman left to return to its non-profit mission and vision, stick to its original values, and adjust the boundaries between non-profit organizations and commercial companies?
Will OpenAI be acquired by Microsoft
There is currently speculation in the industry that with the founder out of the game, in order to avoid possible turmoil, Microsoft will extend an olive branch to OpenAI to seek a takeover.
At present, the biggest financier behind OpenAI is Microsoft. When Microsoft's "money ability" met OpenAI's "superpower", it created the miracle of OpenAI and the miracle of Microsoft's stock price.
Recently, Microsoft's market capitalization has almost surpassed Apple's.
In 2019, Microsoft invested $1 billion in OpenAI and allowed OpenAI to run its models using Microsoft's Azure cloud servers. Because model training is very expensive, OpenAI spends $70 million a year on model training on Microsoft's cloud servers, which is paid for by Microsoft.
In 2023, Microsoft will invest an additional $10 billion in OpenAI, which is probably the largest investment in the history of the artificial intelligence field, and it also represents Microsoft's judgment on the future of the AI field.
Jack Ma once said that "all fields are worth doing all over again with the Internet", and now it is also true that all fields are worth being done by AI.
Microsoft has started "All in AI" internally, and all its products must be embedded with ChatGPT, including Bing, Office Family Bucket (Word, Excel, PPT, respectively), Azure cloud services, etc. This will undoubtedly deepen Microsoft's moat at all levels of the business.
In the future, if Microsoft acquires OpenAI, it will solve OpenAI's tangled value dilemma. Following the "cloud", Microsoft has firmly grasped the core engine of AI. The question is, Altman was expelled for excessive commercialization, will OpenAI's principals still allow it to be acquired by Microsoft?
"Tan Viewpoint" is a Sankei observation column produced by Tencent News and led by Xu Tan, a "media person + writer" - a more professional and interesting interpretation in the uproar.