Sam Altman’s ‘Inconsistent Candor’ Is Exhibiting

When OpenAI’s board fired Sam Altman in late 2023, the board members stated he “was not persistently candid in his communications.” The assertion raised extra questions than solutions, not directly calling Sam Altman a liar, however about what precisely? Six months later, creatives and former workers are as soon as once more asking the general public to query OpenAI’s trustworthiness.

This month, OpenAI claimed ChatGPT’s voice, Sky, by no means meant to resemble Scarlett Johansson from Her, inflicting the award-winning actress to issue a damning public statement and threaten authorized motion. The voice in query is now taken down. Additionally this month, two big names in the AI community who led a security staff inside OpenAI give up. One of many executives, Jan Leike, stated on his manner out that OpenAI’s “security tradition and processes have taken a backseat to shiny merchandise.” As Ed Zitron writes, it’s turning into tougher and tougher to take OpenAI at face worth.

Firstly, the declare that Sky doesn’t sound like Johansson is actually unbelievable. Gizmodo wrote an article claiming it sounded just like the film Her proper after the launch of GPT-4 Omni, as did many different publications. OpenAI executives appeared to jokingly trace on the likeness across the launch. Altman tweeted the phrase “her” on that day. OpenAI’s Audio AGI Analysis Lead has a screenshot from the film as his background on X. All of us might see what OpenAI was going for. Secondly, Johansson claims Altman approached her twice about voicing ChatGPT’s audio assistant. OpenAI says Sky was a special actor altogether, however the declare strikes many as disingenuous.

Final week, Altman stated he was “embarrassed” about not realizing his firm pressured workers to remain quiet about any dangerous experiences at OpenAI for all times or hand over their fairness. The lifelong non-disparagement settlement was revealed by a Vox report, talking to at least one former OpenAI worker who refused to signal it. Whereas many firms have non-disclosure agreements, it’s not on daily basis you see one this excessive.

Altman stated in an interview again in January that he didn’t know whether OpenAI’s Chief Scientist Ilya Sutskever was working on the firm. Simply final week Sutskever and his co-lead of Superalignment, Leike, give up OpenAI. Leike stated Superalignment’s sources had been being siphoned away to different elements of the corporate for months.

In March, Chief Know-how Officer Mira Murati stated she wasn’t sure whether or not Sora was skilled on YouTube movies. Chief Working Officer Brad Lightcap doubled down on this confusion by dodging a question about it at Bloomberg’s Tech Summit in Could. Regardless of that, The New York Times stories that senior members of OpenAI had been concerned in transcribing YouTube movies to coach AI fashions. On Monday, Google CEO Sundar Pichai instructed The Verge that if OpenAI did practice on YouTube movies, that might not be acceptable.

In the end, OpenAI is shrouded in thriller, and the query of Sam Altman’s “inconsistent candor” simply gained’t go away. This can be damaging OpenAI’s status, however the thriller additionally works in OpenAI’s favor. The corporate has painted itself as a secretive startup with the important thing to our futuristic world, and in doing so, OpenAI has efficiently captured our collective consideration. All through the thriller, the corporate has continued to ship cutting-edge AI merchandise. That stated, it’s arduous to not have skepticism concerning the communications from OpenAI, an organization constructed on the premise of being “open.”