OpenAI CEO Sam Altman expressed the view that 'we should accept some bad things in order to reap the benefits of AI.'
by Steve Jurvetson
Sam Altman, CEO of OpenAI, known for developing ChatGPT and other technologies, stated in an interview with Decoded by POLITICO, a newsletter and podcast service run by the news site POLITICO, that when considering the benefits that AI brings, a certain level of risk must be accepted.
Sam Altman to Decoded: 'The world should accept some bad things happening' for the benefits of AI - POLITICO
The interview was conducted by Brendan Borderon, co-author of Decoded and senior technical reporter at POLITICO.
Altman's OpenAI had long advocated for 'looser' regulations on AI. However, following several recent security incidents, including the hacking of Hugging Face by OpenAI's AI agent, they now agree with Anthropic CEO Dario Amodei's view that 'AI development should be slowed down.'
Anthropic CEO Dario Amodi argues that 'AI could take over the internet within 6-12 months' and that 'AI development should be slowed down,' with Elon Musk, Sam Altman, and Demis Hassabis among those who agree - GIGAZINE
Altman's actions seemed unusual, especially since the Wall Street Journal had reported that there had been a decade-long feud between him and Amodi, and that the conflict had yet to be resolved.
However, in an interview with Bauderon, Altman expressed the view that 'there are fundamental differences in worldview (between OpenAI and Anthropic).' When Bauderon then asked, 'Specifically, on what points do you and Amodey disagree?', Altman replied, 'Considering the benefits that AI technology brings and the importance of people remaining autonomous, I believe the world should accept that some negative things may happen as a result of AI.'
Altman acknowledged that he understood the perspective of those who want to strictly regulate AI, such as Anthropic, who argue that 'AI technology is dangerous, so only one company should own it, manage it securely to prevent bad things from happening, and consider how to distribute the benefits.' However, he stated that this is incompatible with OpenAI's proposed 'loose AI regulation.' He said, 'We cannot agree to conditions such as 'guaranteeing that there will be no large-scale hacking, misuse of the technology, fraud, or other wrongdoing.' Because we believe that people will do far more good than bad.' However, he also stated that catastrophic risks, including a 'serious loss of control over AI,' are unacceptable.
Related Posts:
in AI, Posted by logc_nt