Mira Murati (Ex-OpenAI CTO) Launches Thinking Machines Lab
New AI Startup Launched by Former OpenAI CTO
Mira Murati, previously the Chief Technology Officer at OpenAI, has unveiled her new venture. As anticipated, the focus of this startup lies within the realm of artificial intelligence.
Introducing Thinking Machines Lab
Emerging from a period of stealth development, Thinking Machines Lab aims to develop tools that tailor AI to individual requirements and objectives. The company intends to create AI systems that are more accessible, adaptable, and broadly functional than current offerings.
Leadership Team
Murati will serve as the CEO of Thinking Machines Lab. John Schulman, a co-founder of OpenAI, has been appointed as the company’s chief scientist. Barret Zoph, formerly leading model post-training efforts at OpenAI, will assume the role of CTO.
Addressing Existing Gaps in AI
A recent blog post shared with TechCrunch highlights that despite significant advancements in AI capabilities, crucial shortcomings persist.
- The current understanding of cutting-edge AI systems is lagging behind their rapid development.
- Knowledge regarding the training processes of these systems is largely confined to leading research laboratories.
- Despite their potential, these systems often prove difficult to customize for specific needs and values.
Focus on Collaborative and Adaptable Systems
Thinking Machines Lab intends to concentrate on building “multimodal” systems designed for collaborative interaction with users. These systems will be engineered to adapt to diverse human expertise and broaden the scope of potential applications.
Advancing Scientific Discovery
The company is developing models at the forefront of capability in fields such as science and programming. The goal is to unlock transformative applications and benefits, including facilitating novel scientific discoveries and engineering advancements.
Commitment to AI Safety
AI safety is a central principle guiding Thinking Machines Lab’s work. The company plans to contribute to safety through preventing model misuse, sharing best practices for building secure AI systems, and supporting external alignment research by providing code, datasets, and model specifications.
Prioritizing Real-World Value
Thinking Machines Lab will prioritize understanding how its systems generate genuine value in practical applications. The company believes that the most significant breakthroughs often arise from re-evaluating objectives, rather than simply optimizing existing metrics.
Murati’s Transition from OpenAI
Murati departed from OpenAI last October after a six-year tenure, stating her intention to pursue independent exploration.
Previous Roles at OpenAI
Joining OpenAI in 2018 as VP of applied AI and partnerships, Murati was later promoted to CTO in 2022. In this capacity, she oversaw the development of ChatGPT, DALL-E, and Codex, which underpinned early iterations of GitHub’s Copilot.
Brief Tenure as Interim CEO
Murati briefly served as OpenAI’s interim CEO following the unexpected dismissal of CEO Sam Altman, who has described her as a close colleague.
Building a Talented Team
Rumors circulated for months regarding Murati’s recruitment of prominent AI researchers and personnel for her new venture. Thinking Machines Lab’s website lists 29 employees originating from OpenAI, Character AI, and Google DeepMind, among other leading organizations.
Current Hiring Efforts
The company is currently seeking machine learning scientists, engineers, and a research program manager.
Potential Funding
Reports suggested Murati was in discussions to secure over $100 million in funding from undisclosed venture capital firms, though the blog post did not confirm these details.
Prior Experience
Before joining OpenAI, Murati spent three years at Tesla as a senior product manager for the Model X, contributing to the early development of Autopilot. She also held the position of VP of product and engineering at Leap Motion, a startup specializing in hand- and finger-tracking motion sensors.
Joining a Growing Trend
Murati is now part of a growing number of former OpenAI executives launching their own startups, including competitors like Ilya Sutskever’s Safe Superintelligence and Anthropic.