William Saunders, a former OpenAI employee and member of the company’s “superalignment team,” recently disclosed that he quit the company because he felt it was on a collision course with tragedy, much like the ill-fated R.M.S. Titanic in 1912.
As spotted by Business Insider, Saunders gave his commentary during an episode of technology journalist Alex Kantrowitz’s podcast.
Saunders told Kantrowitz that during his three years at OpenAI, he would sometimes ask himself if the company was on a path that was “more like the Apollo program or more like the Titanic.“
His reasoning appears to be that through cautious scientific effort, the Apollo space program was a success despite its setbacks, whereas the Titanic was a failure despite its successes.
Per Saunders:
“Lots of work went into making the ship safe and building watertight compartments so that they could say that it was unsinkable. But at the same time, there weren’t enough lifeboats for everyone.”
OpenAI as the Titanic
Saunders describes the company’s general ethos as one that places a higher priority on product development than on implementing reasonable safety precautions.
He paints a picture of an office where employees who raise safety concerns are punished, and leadership is focused on profit.
“Over time,” said Saunders, “it started to really feel like the decisions being made by leadership were more like the White Star Line building the Titanic.”
The iceberg
Saunders isn’t the first person to disagree with the way OpenAI is allegedly doing business. Elon Musk, one of the company’s estranged co-founders, recently sued it over claims it had strayed too far from its original mission. This is despite the fact that Musk hasn’t had any involvement with OpenAI for years and currently operates his own direct competitor, xAI. The suit was subsequently withdrawn.
Rival AI company Anthropic was founded in 2021 by former OpenAI employees and co-founders over concerns that OpenAI wasn’t focused enough on trust and safety.
On the flip side, though, Ilya Sutskever, another now estranged co-founder and its former chief scientist, departed the company in May 2024 to start his own company, writing at the time that he was “confident” the OpenAI team would be able to safely meet its goals.
With so many employees jumping ship, and the Titanic analogy being bandied about by at least one, the question at hand is: Who or what is the iceberg?
If the passengers in the analogy are us — all humankind — then the iceberg represents the existential threat that ChatGPT and systems like it pose to an unprepared society. If AI becomes more intelligent and capable than us, and we don’t have enough “lifeboats” to protect everyone from its potential for harm, then we could be headed for tragedy.
But if the passengers are shareholders and stakeholders waiting for OpenAI to create AI that’s as smart as humans, and that never happens, then perhaps the iceberg represents the bursting of the generative AI bubble.
Related: Big Tech faces financial reckoning if human-level AI doesn’t happen soon