If you've spent the last two years quietly panicking that AI is coming for your job, your hobbies, and eventually your entire personality, here's someone who might make you feel slightly better about the robot apocalypse.
Mira Murati - former CTO of OpenAI and the person who arguably helped steer ChatGPT into the cultural phenomenon it became - has a new venture called Thinking Machines Lab. And according to an interview with WIRED, her whole pitch is refreshingly un-terrifying: she wants to build AI that keeps humans in the loop, not AI that boots humans out of the loop entirely and changes the locks.
Collaboration, not replacement (allegedly)
In an industry that sometimes seems to be speedrunning "replace all human workers," Murati's stated philosophy is almost quaint. She isn't interested in automating people out of their jobs. She wants AI that can genuinely collaborate - systems that work alongside humans rather than instead of them.
Now, cynics among us might raise an eyebrow. "Keeping humans in the loop" has become something of a buzzword in AI circles, deployed liberally by companies that then proceed to automate entire departments. But coming from Murati, who has more firsthand knowledge of frontier AI development than basically anyone on the planet, it carries a bit more weight than your average press release talking point.
Why this actually matters
The framing here is significant. There are two dominant schools of thought in AI development right now: the "AGI or bust" crowd racing toward full autonomy, and a quieter camp that thinks the most valuable AI might be the kind that makes humans dramatically better at what they already do.
Murati seems to be pitching Thinking Machines Lab firmly in the second camp. And if she can execute on that vision, it's actually a more interesting product design challenge than just building something smarter and hoping for the best. Collaborative AI requires understanding human workflows, human judgment, and human messiness in ways that pure automation doesn't.
The big question
Of course, the gap between "we want to keep humans in the loop" and actually building systems that do that meaningfully is enormous. Intentions are cheap; architecture is hard. But the fact that someone with Murati's credentials and insider knowledge is publicly staking out this position - at a time when the loudest voices in AI are all about autonomy and speed - is at least worth paying attention to.
Whether Thinking Machines Lab delivers on the promise remains to be seen. But in a landscape full of "move fast and automate everything," a little "slow down and collaborate" energy is genuinely welcome.
