Mira Murati Wants Her AI to ‘Keep Humans in the Loop’

Mira Murati, founder and CEO of Thinking Machines Lab and former CTO of OpenAI, is charting a different course for the future of artificial intelligence. Unlike many in the field focused on automation and replacing human jobs, Murati is committed to designing AI that actively collaborates with people.

A Collaborative Approach to AI

Murati believes the journey to AI superintelligence should prioritize human involvement at every step. She envisions a future where individuals can customize powerful AI models to work alongside them, rather than be sidelined by automation. "The best way to ensure positive outcomes with AI is to keep humans in the loop for as long as possible," says Murati.

Introducing Interaction Models

This philosophy took shape with the recent unveiling of Thinking Machines' "interaction models." These advanced systems are trained to interpret natural, unscripted human communication through audio and video, not just text. Unlike typical voice interfaces, these models adapt dynamically to the varied rhythms of conversation, understanding pauses, interruptions, and shifts in tone.

While these capabilities have been demonstrated through internal videos, the full models are not yet available to the public.

Differentiating from Industry Trends

In contrast to major players like OpenAI, Anthropic, and Google—who are pushing toward highly autonomous AI—Thinking Machines is part of a new wave of startups, such as Humans&, dedicated to creating technologies that emphasize human partnership and personalization.

Empowering Users with Custom AI

The company's first product, Tinker, launched in late 2025. It lets researchers and engineers fine-tune frontier AI models with their own data via a user-friendly API, empowering users to customize AI for their unique needs. Multimodal AI expert Alexander Kirillov notes that these new interaction models "constantly perceive what you're doing and can respond, assist, or find information in ways current systems cannot."

A Vision for Amplifying Human Potential

Murati sees these early products and research as foundational steps in building AI that amplifies individual values and preferences, with technology tuned to better understand and anticipate human intent.

For the full original article by Will Knight, visit WIRED.