ChatGPT will be able to talk to you like Scarlett Johansson in Her

OpenAI is releasing a Her-inspired voice assistant feature that can read your facial expressions and translate spoken language in real time — and hopefully do it all without abandoning you like in the movie.

During a livestream demonstration on Monday, OpenAI engineers and CTO Mira Murati gathered around a phone to show the new capabilities. They encouraged the assistant to be more expressive while making up a bedtime story, then abruptly requested it to switch to a robotic voice, before finally asking it to conclude the story with a singing voice. Later, they asked the assistant to look at what the phone’s camera is seeing and have it respond to what’s visible on-screen. The assistant was also able to be interrupted while speaking and respond without continued prompting while acting as a translator.

The assistant’s voice response bore a striking resemblance to the character Scarlett Johansson plays in the movie Her, where a man forms a relationship with a sophisticated AI assistant. After the event, OpenAI CEO Sam Altman cryptically posted just one word on X: “her.” He has also expressed that Her is his favorite movie. The film explores themes of loneliness and human-AI relationships; it seems unlikely that director Spike Jonze intended for the world to precisely replicate that sense of robotic isolation.

In a briefing with The Verge, Murati said that the assistant is not actually designed to sound like Johansson and emphasized that OpenAI has had these voices for a while. “Someone asked me in the audience this exact same question, and then she said, ‘Ah, maybe the reason I didn’t recognize it from ChatGPT is because the voice has so much personality and tonality,’” Murati said.

These features represent a substantial upgrade over ChatGPT’s existing voice mode, which can chat with a user but with much more limited interaction; the current version can’t be interrupted or respond to what your camera sees, for instance. The new capabilities will launch in a limited “alpha” release in “the coming weeks” and be available to ChatGPT Plus subscribers first once a wider rollout begins.

The new voice assistant comes on the heels of a Bloomberg report that claims OpenAI is nearing a deal with Apple to put ChatGPT on the iPhone. (When asked in the briefing, Murati said, “We haven’t talked about any of the partnerships.”) The iPhone voice assistant, Siri, is notoriously unreliable, so a Her-inspired assistant baked into the iPhone that may actually be able to answer your questions instead of “searching the web” seems to be where this is headed.

“The new voice (and video) mode is the best computer interface I’ve ever used. It feels like AI from the movies; and it’s still a bit surprising to me that it’s real,” Altman said in a blog post just after the livestream. “Getting to human-level response times and expressiveness turns out to be a big change.”