Mira Murati, former CTO of OpenAI and current CEO of Thinking Machines Lab, gave an interview to Bloomberg in San Francisco. This was her first major public appearance in about 18 months. For the past year or so, the company has kept a low profile, focusing primarily on fundraising, recruiting researchers, and launching Tinker, an API product for fine-tuning open-source models.
Interaction models become a key focus for external communication.
Murati didn't announce many new plans this time, but she did provide a more systematic overview of the "interaction model" the company is developing. According to her, this model is not a traditional question-and-answer interface, but rather continuously processes audio, text, and video input at intervals of approximately 200 milliseconds.
She stated that this design aims to more closely resemble real human communication, including interruptions, pauses, and mid-expression corrections. She described this direction as the first step in the company's product roadmap, rather than a completed, mature product. She did not provide a specific release date.
Responding to the OpenAI board controversy
In the interview, Murati also discussed the controversy surrounding the removal of Sam Altman by the OpenAI board in November 2023, during which she briefly served as interim CEO. She stated that her judgment at the time revolved around two points: protecting the company's mission and stabilizing the team.
She said that without her involvement during those days and in the subsequent phases, OpenAI might have "collapsed." However, she also acknowledged that, in hindsight, she should have been more proactive in demanding more information and pushing for a clearer transition and greater transparency. She did not directly answer whether she still trusts Altman.
Shifting focus to industry governance
In the interview, Murati repeatedly steered the conversation toward broader industry issues. She believes that what is truly alarming is not a particular individual leader, but rather the excessive concentration of key decisions in the hands of a few people, coupled with a lack of sufficient checks and balances.
She argues that the industry has historically focused too much on the personal qualities of leaders and not enough on governance structures. She worries that this situation isn't unique to OpenAI, but exists across the broader AI industry.
On talent mobility and competitive environment
Regarding the recent departures of several prominent researchers from Thinking Machines, Murati attempted to downplay the impact. She stated that building a cutting-edge AI lab from scratch would compress organizational upheavals that would normally take years into a few months.
She also mentioned that the high salaries in the AI talent war do attract attention, but this is usually not the only reason for staff turnover. She did not position the company with "beating the competition" as its core goal, but rather emphasized product direction and research pace.
Maintain a cautious stance on the future of AI.
In discussions about AI's impact on employment, social risks, and potential misuse, Murati does not accept the binary narrative of "inevitably leading to utopia" or "inevitably leading to loss of control." She believes that the current stage is still shaping the future direction of AI.
However, she also emphasized that if humans prematurely loosen their control over the system, the future outcomes could be significantly different, and not necessarily better. This statement echoes her repeated mentions of governance and human intervention in the interview.
