Mira Moratti wants her AI to ‘keep humans informed’

Mira Moratti still He wants to build super artificial intelligence. But the former CTO of OpenAI sees human intelligence as an important part of the equation.

At a time of growing concern about AI eliminating jobs and increasing the power of a few big companies, Moratti’s startup, Thinking Machines Lab, offers a radically different vision of the technology.

“At some point, we’re going to have superintelligent machines,” Moratti tells WIRED. “But we think the best way to have many potential futures — that is, good futures — is to keep humans informed.”

Moratti says AI doesn’t need to automate humans out of the equation. She suggests that a more optimistic approach is to let people build and customize their own leading AI models, and then work with those models to achieve their goals.

Thinking Machines this week previewed a new type of AI model that it says points to a more inclusive reality for humans. The company’s “interaction models” are trained to communicate with someone through a camera and microphone. Unlike many existing voice-mode interfaces, the new models don’t simply capture speech, transcribe it, and then feed it into a language model that processes it in the same way as a chatbot. Interaction models fundamentally understand continuous, chaotic human communication, which means they are better able to understand the meaning of pauses, interruptions, and changes in tone. This allows them to adapt quickly when someone makes a point or changes the topic. The company has shown several videos demonstrating these capabilities, although models have not been released publicly.

Moratti’s approach contrasts with the way most major AI companies seem to be pursuing superintelligence today. OpenAI, Anthropic, and Google are developing large models that do increasingly complex work, including writing entire software applications from scratch, via text message. This requires a little help from a human.

Thinking Machines isn’t the only startup envisioning a more inclusive future for humans. Other labs, including Humans&, also aim to develop AI systems that prioritize human collaboration. Some prominent economists have called for AI researchers and companies to build systems in this way, with a focus on empowering humans rather than replacing them.

Moratti left her position as CTO at OpenAI in 2024, and co-founded Thinking Machines with several prominent engineers. Thinking Machines has raised billions of dollars to build frontier AI.

However, the company has so far only released one product. Tinker, which launched in October 2025, enables frontier AI model optimization using custom data. It is available today as an Application Programming Interface (API) that researchers and engineers can use to tune open source models.

The lab’s new interaction models also have the potential to enable more personalized and customized AI, says Alexander Kirillov, a founding member of the Thinking Machines team and an expert in multimodal AI, that is, models that handle audio and video as well as text.

“The model is constantly aware of what you are doing, and it is always there to respond to you and provide you with information, search for information, or use other tools,” says Kirillov. “This is something no one has [today’s other] Models can actually do. Turns [in a conversation] They are determined by a much less intelligent system.

This is all part of a larger vision for artificial intelligence, Mira says.

“This shows the first bet on human cooperation,” she explains. “Where this is going is amplifying people’s preferences and values, where AI understands intentions and predicts intentions.”