Stopping AI’s Runaway Dangers Will Take More Than Just Talk About P(doom)
Welcome to CNET’s series of guest columns called Alt View, a forum for a diverse array of experts and luminaries to share their insights into the rapidly evolving field of artificial intelligence. For more AI coverage, check out CNET’s AI Atlas.
I doubt anyone working in AI was surprised when Jacob Coxon resigned from Anthropic earlier this month, warning of an impending apocalypse. Nor when Evan Hubinger, who runs Anthropic’s “alignment” research agreed with Coxon, saying he reckons the odds of AI killing all humans within the next decade is over 10% (and he doesn’t have a plan to fix it).
They’ve heard all this before: Many inside the leading labs talk openly about their p(doom) score — the presumed likelihood of an AI-induced doomsday event. It’s even possible Evan thought we’d be relieved. Only 10%! Other industry insiders put the odds significantly higher.
We weren’t relieved, though, were...
Copyright of this story solely belongs to www.cnet.com. To see the full text click HERE
