2024/08/06 12:36
OpenAI's top management has changed dramatically: Lianchuang resigned and the president went on leave, netizens: The leadership has become an empty shell
2024/08/06 12:36
Just now, OpenAI high-rise earthquake:
John Schulman resigned and ran away, Greg Brockman, president of Lianchuang & Greg Brockman, took a long vacation, and Peter Deng, vice president of product, was also exposed.
John ·Schulman, who has led the OpenAI reinforcement learning team for many years, is known as the "ChatGPT architect".
It was also after the departure of Ilya Sustkever that he temporarily took over the Super Alignment team.
Now he's announcing that he has "made a difficult decision" to join Anthropic next door, aka the Claude team.
At the same time, Greg Brockman, president of OpenAI, also announced that he would take a long vacation until the end of the year.
OpenAI plans to hold events around the world such as developer days from October 1, and he will also be absent.
Brockman said that he relaxed for the first time after 9 years of continuous work, but he was vague about his plans after the end of the year, and some netizens thought that it was due to a non-disclosure agreement.
Twenty minutes after Schumann's departure was announced, Ultraman thanked Schumann for his contribution in his reply, as well as recalled the first time the two met in 2015.
But as of now, Altman has not responded to Brockman's leave.
Ultraman usually has a habit of typing the first letter of a sentence and "i" without capitalization, but whenever an important colleague chooses to leave, he will behave more formally.
Brother Apple, a well-known OpenAI insider breaking news account, said: Now that the leadership has become an empty shell, you better hurry up and release the product.
Schumann studied physics at Caltech, neuroscience at UC Berkeley, and completed his Ph.D. in computer science under the tutelage of reinforcement learning guru Pieter Abbeel.
When he became the co-founder of OpenAI, he had not yet graduated with a Ph.D.
OpenAI is the first and only company he's worked for outside of his internship.
His masterpiece and most cited paper PPO is the reinforcement learning algorithm used in ChatGPT's core technology RLHF.
Later, he led the alignment/post-training team in a series of work from GPT-3.5, GPT-4 to GPT-4o.
The reason for his resignation this time is also related to this:
Hopefully deepening my focus on AI alignment and starting a new chapter in my career where I can get back to working in real technology.
However, he also gave OpenAI enough face, adding that "it was not because of OpenAI's lack of support for security alignment efforts that he left", but just out of a personal desire to concentrate.
In his ICML2023 talk, Schumann described one of his next research directions: how to avoid over-optimization.
In an interview with podcast host Dwarkesh Patel in May, Schumann also said that computing power should shift more from pre-training to post-training.
Due to the disagreement between the concept of AI security and Ultraman, OpenAI has experienced a large loss of talents.
In May, two leaders of the Super Alignment team, Chief Scientist Ilya Sutskever and Jan Leike, one of the inventors of RLHF, left OpenAI on the same day.
At that time, it was less than a year before the establishment of the Super Alignment team.
After Jan Leike left, he posted more than a dozen tweets in a row, bombarding OpenAI, accusing it of not fulfilling its promise of 20% computing power, and paying far less attention to security than its products.
Later, Jan Leike chose to join OpenAI's rival Anthropic to continue his research on super-alignment.
Ryan Lowe, who co-led the alignment with Jan Leike on the GPT-4 project, also left OpenAI at the end of March.
William Saunders, one of the members of the Super Alignment, and Daniel Kokotajlo, who participated in the GPT-4 adversarial test, also left OpenAI.
Not to mention the siblings Dario Amodei and Daniela Amodei, the founders of Anthropic next door, are also from OpenAI.
His older brother Dario was the vice president of research at OpenAI before leaving, and his younger sister Daniela was the deputy director of security and strategy, and she brought more than a dozen employees including Tom Brown, the chief engineer of GPT-3, to Anthropic.
One of the reasons why they chose to leave the business was that OpenAI directly released GPT-3 without the security issue being resolved, which caused their dissatisfaction.
The emphasis on safety, combined with the founder's background, has also made Anthropic one of the main places to go for OpenAI employees, including Jan Leike and Schumann who left the company.
OpenAI has also recently made some moves in security work-
At the end of July, OpenAI moved Aleksander Madry, a senior director of security, from his security position and reassigned him to "AI inference."
However, OpenAI told the media that Madry will still work on AI safety after taking up his new position.
CEO Altman also emphasized that OpenAI has always attached great importance to security and adhered to the promise of "giving the entire security team at least 20% computing power".
But sharp-eyed netizens soon discovered that Ultraman played a wave of stealing concepts, and the original promise was to give the super alignment team 20% of the computing power, which was secretly changed to all security teams.
Therefore, after Ultraman's tweet, netizens also added this background information mercilessly, and attached the official link of OpenAI at that time.
emm…… The Internet still has a memory.
While a large number of internal talent is draining, OpenAI's "old enemy" Musk also has new actions -
Musk has just taken OpenAI to court again, along with Ultraman and President Brockman.
The reason for this lawsuit is that Musk believes that he was "deceived and blackmailed" by Ultraman and Brockman when he participated in the creation of OpenAI.
Musk said that he participated in and invested in OpenAI because he believed Altman's words, and Altman promised that OpenAI was a safe and open non-profit organization.
But then OpenAI became a for-profit organization and sold a part of its shares to Microsoft, which made Musk feel very dissatisfied, believing that he had been betrayed and denounced Ultraman for "treachery".
Musk's lawyer, Marc Toberoff, said the lawsuit was a big difference from the one previously withdrawn, holding the defendants "accountable for knowingly making false statements to Musk and the public" and seeking to "recover their ill-gotten gains on a large scale."
At the same time, the indictment also seeks to revoke the license granted by OpenAI to Microsoft.
Reference link:
[1]https://www.bloomberg.com/news/articles/2024-08-06/openai-co-founder-john-schulman-departs-for-ai-rival-anthropic
[2]https://www.cnbc.com/2024/07/23/openai-removes-ai-safety-executive-aleksander-madry-from-role.html
[3]https://x.com/sama/status/1818867964369928387
[4]https://www.documentcloud.org/documents/25031741-elon-vs-openai-again
[5]https://www.nytimes.com/2024/08/05/technology/elon-musk-openai-lawsuit.html
This article is from Xinzhi self-media and does not represent the views and positions of Business Xinzhi.If there is any suspicion of infringement, please contact the administrator of the Business News Platform.Contact: system@shangyexinzhi.com