New York
CNN
—
OpenAI says it’s hitting the pause button on an artificial voice launched with an update to ChatGPT that prompted comparisons with a fictional voice assistant portrayed within the quasi-dystopian movie “Her” by actor Scarlett Johansson.
The retreat by OpenAI follows a backlash to the synthetic voice, generally known as Sky, which critics described as being overly aware of customers and sounded as if it had emerged from a male developer’s fantasy. It was broadly mocked for its flirtatious tone.
“We’ve heard questions on how we selected the voices in ChatGPT, particularly Sky,” OpenAI mentioned in a post on X Monday. “We’re working to pause using Sky whereas we tackle them.”
The voice in query is just not derived from Johansson’s, the corporate mentioned in a blog post Sunday, however as a substitute “belongs to a unique skilled actress utilizing her personal pure talking voice.”
OpenAI mentioned that with every of its AI voices, it tried to create “an approachable voice that conjures up belief,” one which incorporates a “wealthy tone” and is “pure and simple to hearken to.” The ChatGPT voice mode that used the Sky voice had not but been broadly launched, however movies from the product announcement and teasers of OpenAI staff talking with it went viral on-line final week.
Some who heard Sky derided it as maybe too simple to hearken to. Final week, the controversy impressed a segment on The Daily Show through which senior correspondent Desi Lydic described Sky as a “attractive robotic child voice.”
“That is clearly programmed to feed dudes’ egos,” Lydic mentioned. “You’ll be able to actually inform {that a} man constructed this tech.”
Even OpenAI CEO Sam Altman appeared to acknowledge the widespread parallels customers had been drawing with Johansson when he posted to X on the day of the product’s announcement: “her.”
“Her” is the title of the 2013 movie through which Johansson portrays an artificially clever voice assistant with whom the protagonist, performed by Joaquin Phoenix, falls in love, solely to be left heartbroken when the AI admits she can be in love with a whole bunch of different customers and later turns into inaccessible altogether.
The criticism surrounding Sky highlights broader societal considerations in regards to the potential biases of a know-how designed by tech firms largely led or funded by White males.
The announcement got here after OpenAI leaders had been compelled to defend their security practices over the weekend after a departing worker called the company’s priorities into query.
Jan Leike, who previously led a crew centered on long-term AI security however left OpenAI final week together with Co-Founder and Chief Scientist Ilya Sutskever, posted a thread on X Friday claiming that “over the previous years, security tradition and processes have taken a backseat to shiny merchandise” at OpenAI. He additionally raised considerations that the corporate was not devoting sufficient sources to making ready for a attainable future “synthetic common intelligence” (AGI) that may very well be smarter than people.
Altman rapidly responded saying he appreciated Leike’s dedication to “security tradition” and added: “He’s proper we have now much more to do; we’re dedicated to doing it.” The corporate additionally confirmed to CNN that in current weeks it had begun to dissolve the crew Leike led, and as a substitute was integrating members of the crew throughout its numerous analysis teams. A spokesperson for the corporate mentioned that construction would assist OpenAI higher obtain its security aims.
OpenAI President Greg Brockman responded in a longer post on Saturday, which was signed with each his identify and Altman’s, laying out the corporate’s strategy to long-term AI security.
“We’ve got raised consciousness of the dangers and alternatives of AGI in order that the world can higher put together for it,” Brockman mentioned. “We’ve repeatedly demonstrated the unimaginable potentialities from scaling up deep studying and analyzed their implications; known as for worldwide governance of AGI earlier than such calls had been standard; and helped pioneer the science of assessing AI programs for catastrophic dangers.”
He added that as AI turns into smarter and extra built-in with people’ day by day lives, the corporate is targeted on having in place “a really tight suggestions loop, rigorous testing, cautious consideration at each step, world-class safety, and concord of security and capabilities.”