Mira Murati, former OpenAI chief technology officer, launched Thinking Machines Lab on Tuesday: a new AI startup aimed at tackling the alignment problem by encoding human values into artificial intelligence systems.
The company, staffed with roughly 30 leading researchers and engineers, includes about 20 poached from OpenAI, alongside talent from Meta and Mistral. Murati will serve as CEO, leveraging her experience as a key architect of ChatGPT during her tenure at OpenAI from June 2018 until her abrupt exit in September 2024.
The startup’s mission centers on AI alignment: a process to ensure models are safe, reliable, and reflective of human ethics. In a company blog post, Thinking Machines Lab emphasized building systems that “adapt to the full spectrum of human expertise and enable a broader spectrum of applications,” distinguishing itself from competitors focused on narrower domains like programming or mathematics. To support this goal, the firm commits to sharing code, datasets, and model specifications with the research community.
Thinking Machines Lab enters a competitive landscape alongside Anthropic and Safe Superintelligence—both founded by ex-OpenAI leaders—but its focus on human-value alignment offers a unique angle. My question about this is simple: since I can’t find two human beings whose human values align, whose human values are they going to align their models with?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.