On Tuesday (February 18) local time, former OpenAI executive Mira Murati announced the establishment of a new artificial intelligence company, Thinking Machines Lab, on social media X. The official website writes that the company will focus on building artificial intelligence (AI) models and products to support more "human-AI collaboration" across work areas. "While the current system is good at programming and mathematics, we are building artificial intelligence that can adapt to all human expertise and achieve wider applications."
The official website posted a list of the company’s 29-person team, more than 20 of whom have experience working in OpenAI. One of the more well-known is OpenAI co-founder John Schulman, who is serving as the chief scientist of the new company.
Barret Zoph, who was the vice president of research at OpenAI, is serving as the chief technology officer of ThinkingMachinesLab; Lilian Weng, the vice president of research who was responsible for security at OpenAI, is also on the team list of the new company.
Mulati once worked at Tesla. She joined OpenAI in 2018 and was promoted to the company's chief technology officer in 2022. When Sam Altman (Sam Altman) was dismissed, she briefly served as OpenAI's CEO and pushed for Altman's reinstatement.
Mulati & Altman
In more than 6 years at OpenAI, Murati has led transformative projects such as the chatbot ChatGPT and the image generation model DALL-E. In September last year, Mulati resigned suddenly. It is reported that Mulati has been secretly preparing to start a business after leaving OpenAI.
Mulati wrote in his latest post that ThinkingMachinesLab is "building three things": ① Helping people adapt artificial intelligence systems to meet their specific needs; ② Laying a solid foundation to build more powerful artificial intelligence systems; ③ Cultivating a culture of open science to help the entire field understand and improve these systems.
ThinkingMachinesLab plans to publish frequent technical blog posts, papers, and code. “Our goal is simple: to advance the development of artificial intelligence by making it broadly useful and understandable through solid foundations, open science, and practical applications,” Mulati said.
Although ThinkingMachinesLab has not yet launched a product or model, it claims to have a different philosophy than some other AI companies: the company is letting researchers and product leaders collaborate to "co-design" to "make artificial intelligence systems more widely understood, customizable and universally capable."
It is worth mentioning that in addition to Mulati, many former OpenAI executives also chose to "leave and work alone."
It was reported yesterday that SafeSuperintelligence, founded by OpenAI co-founder and former chief scientist Ilya Sutskever, is expected to raise more than US$1 billion in the latest round of financing, with a valuation of more than US$30 billion.