Sam Altman debated the safety of AI with an Oxford University student who protested his speech
Sven Hoppe/picture alliance via Getty Images
- Sam Altman gave a speech to an Oxford University student business society in May.
- A group of students protested the event and the OpenAI CEO went to chat with them afterward.
- Altman said that building stronger AI systems is the best way to ensure safety.
When Sam Altman gave a speech to an Oxford University business society in the UK last month, a group of students protested the event – and the OpenAI CEO ended up debating one of them, The Guardian reported.
They discussed concerns over the safety of artificial general intelligence or AGI – the point at which a machine can do anything a person can.
300 million jobs could be impacted by AI according to a Goldman Sachs report, while the technology's leading figures – including Altman – last month signed a statement saying AGI has risks comparable to nuclear war or a pandemic.
According to the newspaper, the student protestors held signs calling on the ChatGPT creator to "stop the AGI suicide race."
After Altman finished his speech, he went over to talk to them.
"Stop trying to build an AGI and start trying to make sure that AI systems can be safe," one of the students told him, per The Guardian.
"If we, and I think you, think that AGI systems can be significantly dangerous, I don't understand why we should be taking the risk," he added.
In a conversation with The New York Times, Altman previously compared his ambitions for OpenAI with the Manhattan Project – codename for the US government's project to produce the first nuclear bomb in World War II.
"Technology happens because it is possible," Altman then said – paraphrasing a Robert Oppenheimer speech where he justified creating the nuclear bombs as a necessary expansion of human knowledge. Oppenheimer led the Manhattan Project.
He echoed similar ideas when replying to the protestor, saying: "I think a race towards AGI is a bad thing, and I think not making safety progress is a bad thing," per The Guardian.
Altman said the only way to achieve safety is with "capability progress," meaning building stronger AI systems in order to understand how they work – even if that's risky, The Guardian reported.
"It's good to have these conversations," he told the reporter afterward.
OpenAI did not immediately respond to Insider's request for comment, sent outside US working hours.