Mira Murati steps back into the spotlight, carefully

Murati Previews 'Interaction Models' in Cautious Return

After 18 months of relative quiet, Mira Murati, CEO of Thinking Machines Lab and former CTO of OpenAI, stepped back into the public sphere with a carefully managed interview at a Bloomberg event. Murati used the appearance to introduce a new concept called “interaction models,” a different approach to AI interfaces, signaling her company is ready to compete for attention in a market dominated by her former employer, Anthropic, and the increasingly visible xAI. The move suggests a strategic shift from a heads-down development phase to a more public-facing effort to carve out a distinct space in the AI landscape.

A New Interface Paradigm

While Murati remained guarded on specific product timelines, she detailed the thinking behind the company's interaction models. Unlike the turn-based, prompt-and-response structure common today, this new interface is designed for more fluid, real-time engagement. The goal is for the AI to process human communication with more nuance, a step beyond what is offered by the lab's current fine-tuning API, Tinker. The key attributes of this approach include:

• Processing continuous streams of audio, text, and video.
• Operating in rapid, 200-millisecond intervals.
• Capturing conversational textures like interruptions and pauses.
• Moving toward a more naturalistic, real-time collaborative tool.

Governance and Concentration of Power

Beyond technology, Murati addressed her brief, chaotic tenure as interim CEO of OpenAI during the November 2023 leadership crisis. She steered the conversation toward a broader industry concern: the concentration of consequential decisions in the hands of too few people. Murati argued that the industry has focused heavily on the virtue of individual leaders while neglecting the need for robust structural checks and governance. This critique positions Thinking Machines Lab not just as a technical competitor but as an organization potentially founded on a different set of principles regarding corporate structure and ethical oversight in AI development.

Murati is signaling that the next competitive frontier isn't just model performance, but the interface paradigm itself. By focusing on real-time 'interaction models' and publicly questioning the industry's governance structures, she is attempting to differentiate Thinking Machines Lab on both technology and principle, a difficult but necessary strategy in a market saturated with well-funded incumbents.