OpenAI's head of alignment, John Schulman, who was responsible for tailoring its model responses to align with values, has left the company to work with Amazon backed Anthropic. Anthropic and OpenAI have a shared history, and the former comprises former executives from the latter who decided to set up their own company. Schulman announced his decision on X, and he was met with responses from OpenAI chief Sam Altman and others who appreciated the time they had spent together working at the company.
Another OpenAI Safety Lead Departs Firm For Amazon's Anthropic
Schulman's X post comes just after a report from The Information claimed that OpenAI's president, Greg Brockman, had taken an extended leave while a product manager, Peter Deng, was also leaving the company. Schulman, who was responsible for overseeing model alignment at OpenAI, will now work at Amazon backed Anthropic. Anthropic was founded by former OpenAI members, and Schulman's departure means that few of OpenAI's founding members now remain at the company—Brockman and Altman being two of them.
As OpenAI's head of alignment, Schulman was responsible for ensuring that the firm's products, such as ChatGPT, produced outputs that were in line with a predefined set of values. Some examples include tailoring responses against widely held beliefs and restricting harmful outputs to ensure that all responses generated by AI models are safe for human use.
In his statement, Schulman shares that he is leaving OpenAI to "deepen" his focus on AI alignment and "return to hands-on technical work." He adds that the decision does not stem from any lack of focus on alignment at OpenAI but is a "personal one."
After his departure, along with Altman and Brockman, the third OpenAI founder currently working at the firm is its head of language and code generation, Wojciech Zaremba.
Responding to Schulman's announcement, OpenAI CEO Altman thanked Schulman for the work he had performed at the company. Altman also shared details of his first meeting with Schulman, revealing that they had met at a cafe at the University of California, Berkley, where they talked about artificial general intelligence.
As opposed to ChatGPT, which is narrow intelligence, artificial general intelligence or AGI is the end goal of all AI firms as it seeks to create software that can deliver multi faceted outputs that mimic the cognitive capabilities of a human being.
Schulman's departure follows that of AI safety researcher Jan Leike in May, Leike left OpenAI after the firm dissolved an alignment team and he joined the Amazon backed Anthropic to lead its alignment efforts. After Leiki left, Schulman took over what OpenAI dubs as 'post training.' His post stressed that post training "is in good hands and has a deep bench of amazing talent" and is "coming together with some promising projects."
I shared the following note with my OpenAI colleagues today:
I've made the difficult decision to leave OpenAI. This choice stems from my desire to deepen my focus on AI alignment, and to start a new chapter of my career where I can return to hands-on technical work. I've decided…
— John Schulman (@johnschulman2) August 6, 2024
About the author: Ramish is a seasoned technology writer and editor with more than a decade of experience. He specializes in semiconductor fabrication and market analysis. With a background in finance and supply chain management - via his bachelors in Finance and a micromasters in supply chain management from MIT - Ramish combines financial rigor with deep industry insight to deliver accurate and authoritative coverage.
Follow Wccftech on Google to get more of our news coverage in your feeds.
