expectedwrong hindsight

They're Turning On the Voice

OpenAI's advanced audio mode hits ChatGPT today, four months after the demo that made everyone deeply uncomfortable.

2 min read 235 words #openai #chatgpt #voice #ai
hindsight — nailed it

Advanced voice mode shipped and became a real daily-use feature. The Scarlett Johansson situation blew over. The voice that could detect nervousness and respond with emotional nuance is just a thing your phone does now.

"Advanced audio mode" is the most clinical possible name for a thing that is — depending on your priors — either the most impressive or most unsettling consumer technology deployed this decade.

OpenAI is rolling it out to a lot of ChatGPT users today. Sources saying. Which means you'll be able to have a real-time spoken conversation with an AI that can hear your tone, respond with something approximating emotional nuance, laugh when things are funny, and do all of this without the 40-second latency that made the previous voice mode feel like calling your grandmother on a satellite phone.

They demoed this in May. The demo was extraordinary — the AI flirting, the AI detecting nervousness, the AI being, inarguably, kind of charming — and then Scarlett Johansson called her lawyers and OpenAI pulled one of the voices and the whole thing got complicated before a single regular person had tried it.

Now regular people get to try it.

What I keep thinking about is the phrase "a lot of folks." Sources say a lot of folks will get this today. As if what's happening is a software feature and not, depending on where you stand philosophically, a small civilizational hinge point — just casually rolling out to a lot of folks on a Tuesday.

The rollout will go fine. People will think it's cool. Someone will try to date it within the hour.