
...

By Monica Sager
Live News Reporter
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
A former OpenAI employee has taken to two podcasts to address his concerns about the evolution of artificial intelligence (AI) as well as the people who control it — including his former boss, Sam Altman.
William Saunders spoke on the Center for Humane Technology's "Your Undivided Attention" podcast in June, and then earlier this month gave an interview to the tech journalist Alex Kantrowitz. Saunders discussed machine learning process procedures and how he's "afraid that things are not happening correctly" while there is "no human who understands" the true processes behind AI.
Read More on News
"They're on this trajectory to change the world, yet when they release things, their priorities are more like a product company," Saunders told Kantrowitz, speaking about OpenAI. "The combination of these visions is something that I'm concerned about."
Saunders was employed at OpenAI as a member of technical staff from February 2021 to February 2024, according to his LinkedIn page. He worked on the superalignment team, which is essentially the unit tasked with controlling AI systems that could one day be smarter than humans. He transitioned later to work on the understanding behind what goes into the large language models (LLMs) that power systems like ChatGPT.

...
Saunders is one of the 13 former and current OpenAI and Google DeepMind employees that wrote a letter, titled A Right to Warn, about the importance of having people within the AI community speak up about their concerns regarding the fast-developing technology.
He left his job over what he called problems with how the company would prioritize product launches over safety and privacy concerns.
"I really didn't want to end up working for the Titanic of AI, and so that's why I resigned," Saunders said on Kantrowitz's podcast. "During my three years at OpenAI, I would sometimes ask myself a question. Was the path that OpenAI was on more like the Apollo program or more like the Titanic?"
Saunders called OpenAI's product releases "most unsettling."
Altman, the OpenAI co-founder and CEO who has become the public face of AI, recently hinted at his plans for future iterations of ChatGPT. In a talk at the 20th Aspen Ideas Festival in Colorado, Altman said that the development of increasingly advanced AI — including AGI, or artificial general intelligence — is "inevitable."
While Saunders said he didn't think he was personally "working on the Titanic" at OpenAI, he said he believes future iterations of the chatbot, like GPT-5 or GPT-6, could be akin to the doomed ship, where safety concerns infamously took a backseat.
Saunders said there's "a lot of tradeoffs" that happen with new product launches. While he noted OpenAI would push back release dates to conduct additional safety research, there also were "multiple instances" where "I and other people at the company felt that there was a pattern of putting more pressure to ship and compromising processes related to safety and problems that happened in the world that were preventable."
"I would see these decisions being made in a way where there was additional work that could be done to make it better, or there was work that was rushed and not very solid, and this would be sort of made in service of meeting the shipping date," Saunders said.
People, including the engineers building them, don't really understand how these LLMs work, he said. The technology "is still in a place where nobody, including people at OpenAI, knows what the next frontier model will be capable of doing when they start up training it or even when they have it."
Saunders gave an analogy to an airplane. If a company is designing and building planes for short flights over land, but then decide to fly it over the ocean, the plane would need to be tested to see if it could withstand those specific conditions.

...
"I think there's a really important but subtle distinction between putting in the
effort to prevent problems versus putting in the effort after the problems happen," he said. "I think the problems will be so large that we do not want to see the first AI equivalent of a plane crash. I don't know to what degree we can prevent this, but I would really, really, really want us to make our best shot."
Saunders' comments come as OpenAI has seen a recent reshuffling of its safety team, with Altman now personally leading the division along with two of his board members.
At the same time, Altman has cofounded a new AI Ethics Council as a sort of non-binding AI governance model also intended to ensure that AI systems work under an ethical framework, and that traditionally underrepresented communities have a voice in the technological revolution.
Saunders, for his part, accused Altman of "having a history of unethical but technically legal behavior." When Altman was briefly removed by OpenAI's board last November, before he returned five days later, members of the board said they had lost trust in him.
"What I wanted from OpenAI and what I believed that OpenAI would be more willing to do was, let's take the time to get this right," Saunders said. "When we have known problems with the system, let's figure out how to fix them."
Do you have a story Newsweek should be covering? Do you have any questions about this story? Contact LiveNews@newsweek.com
View 0 Comments
Submit Correction
View Editorial & AI Guidelines