OpenAI says its new tools mean ChatGPT chatbot can see, hear, speak
OpenAI on Monday said its chatbot has developed humanlike skills needed to converse with people and debuted an overhaul of ChatGPT’s new powers.
The artificial intelligence company has pressed forward with new audio and video tools designed to make its product more lifelike.
OpenAI said that “ChatGPT can now see, hear and speak” and the audio platform Spotify was using the tools to translate podcasts.
“The new voice technology — capable of crafting realistic synthetic voices from just a few seconds of real speech — opens doors to many creative and accessibility-focused applications,” OpenAI said on its blog. “However, these capabilities also present new risks, such as the potential for malicious actors to impersonate public figures or commit fraud.”
OpenAI showcased the beneficial uses of its new offerings on its blog, in videos depicting people using ChatGPT to generate a bedtime story and someone receiving advice on adjusting a bicycle seat from messaging an image of the problem.
Amid concerns about how the audio and visual tools may enable fraud, OpenAI said it would gradually deploy the capabilities and carefully share the offerings with users and developers.
Spotify, the music and podcasting platform, already has deployed a product using the technology.
The audio company said it’s piloting Voice Translation for its podcasts whereby listeners can hear broadcasts in a different language than the original recording.
Spotify is debuting the technology on a handful of podcasts, which will let people listen to English-language episodes in different tongues while keeping the speaker’s distinctive characteristics, such as pacing and syntax.
“By matching the creator’s own voice, Voice Translation gives listeners around the world the power to discover and be inspired by new podcasters in a more authentic way than ever before,” said Ziad Sultan, Spotify vice president of personalization, in a statement.
AI tools that let people impersonate someone’s image and voice already exist.
OpenAI’s Ilya Sutskever and Jan Leike warned in July that AI surpassing human intelligence could arrive before the end of the decade and lead to “human extinction.”
In response to the doomsday concern, OpenAI said it was assembling a team of researchers to help delay the apocalypse.
Not every tech expert has a pessimistic outlook on the arrival of tech tools.
At the Intelligence and National Security Summit in July, Booz Allen Hamilton’s Patrick Biltgen touted the promise of AI assistants to help American intelligence officers solve complex problems.
Mr. Biltgen, who works on AI with U.S. government agencies, said he envisioned spies using AI similar to how the AI character J.A.R.V.I.S. assists Tony Stark in the popular superhero tale “Iron Man.”