A former OpenAI researcher, William Saunders, has resigned, citing concerns that the company prioritises profit over safety.
Saunders likened OpenAI's trajectory to the ill-fated Titanic, raising alarms about the company's approach to developing artificial general intelligence (AGI).
Key Takeaways
- William Saunders, a former member of OpenAI's Superalignment team, resigned due to safety concerns.
- Saunders compared OpenAI's path to the Titanic, suggesting the company prioritises profit over safety.
- OpenAI's CEO, Sam Altman, dissolved the safety-oriented Superalignment team earlier this year.
- Critics argue that OpenAI is more focused on releasing new products and securing funding than ensuring safety.
- The development of AGI remains theoretical but is a significant focus for OpenAI.
The Titanic Analogy
During a recent podcast with tech YouTuber Alex Kantrowitz, Saunders explained his decision to leave OpenAI. He stated, "I really didn't want to end up working for the Titanic of AI, and so that's why I resigned." Saunders questioned whether OpenAI's path was more like the Apollo programme or the Titanic, ultimately concluding it was the latter.
Saunders argued that, like the Titanic, OpenAI was prioritising the release of new, shiny products over safety. He drew parallels between the lack of lifeboats on the Titanic and OpenAI's approach to risk management.
Concerns Over AGI
Saunders' comments highlight growing concerns about companies like OpenAI developing AI systems capable of surpassing human abilities, known as artificial general intelligence (AGI). While AGI remains theoretical, it is a significant focus for OpenAI executives, including CEO Sam Altman.
Several former employees have accused OpenAI leadership of ignoring safety concerns and stifling oversight. Earlier this year, Altman dissolved the safety-oriented Superalignment team, which Saunders was part of, and created a new "safety and security committee."
Shifting Corporate Structure
The dissolution of the Superalignment team and the creation of the new committee have raised questions about OpenAI's commitment to safety. Former chief scientist Ilya Sutskever, who led the now-dissolved team, has since started a new company called Safe Superintelligence Inc. This new venture aims to focus solely on safety in AI development.
Critics argue that despite Altman's claims of prioritising safe AGI development, OpenAI has been more focused on releasing new chatbots, securing major funding rounds, and forming billion-dollar partnerships with tech giants like Microsoft.
A Call for Safety
Saunders hoped OpenAI would operate more like NASA's Apollo space programme, which had redundancies and adaptability to handle significant problems. He acknowledged that developing AGI or any new technology with zero risk is impossible but emphasised the need for taking all reasonable steps to mitigate these risks.
"What I would like to see is the company taking all possible reasonable steps to prevent these risks," Saunders said in an email to Business Insider.
Future Uncertainties
It remains unclear whether the risks Saunders highlighted will materialise. Experts have pointed out that we are still far from developing an AI that can outwit a human, and the technology may not progress as anticipated.
Nonetheless, Saunders' resignation and comments paint a concerning picture of OpenAI's current direction and the changes implemented by Altman over the past few years.
Sources
- OpenAI Researcher Says He Quit When He Realized the Upsetting Truth, Futurism.
- OpenAI Researcher Says He Quit When He Realized the Upsetting Truth, Yahoo News UK.
