Thinking Machines wants to create an AI that listens while it talks

Thinking Machines Lab, an AI startup founded last year by OpenAI CTO Mira Murati, on Monday announced the so-called. related exampleswhich, at best, sounds like an AI that will mess with you.

Right now, every type of AI you’ve ever used works the same way. You talk, it listens. It responds, you listen. Thinking Machines is trying to change this by creating a model that processes your input and creates an instant response, so it’s more like a phone call than a text message.

The technical term for this is “full duplex,” and the company says its model, TML-Interaction-Small, responds in 0.40 seconds, which is about the speed of natural human speech and much faster than similar models from OpenAI and Google.

However, this is a research preview, not a commercial. The company is not yet releasing it to the public. “Small scale research” is coming in the next few months, it says, with a larger rollout expected later this year.

So what to do? We don’t know. The signs it’s interesting and the basic idea – that interaction should be modeled, not built – is interesting. Whether real-world experience matches what experts say is something we won’t know until people actually use it.