Various figures in the tech industry have shared their p(doom) scores, with numbers ranging widely. For instance, Anthropic CEO Dario Amodei has put his p(doom) at 10 to 25, while Eliezer Yudkowsky, cofounder of Berkeley’s Machine Intelligence Research Institute, says his number is higher than 95. Despite the widespread use of the term, there is no consensus on the actual likelihood of AI causing a doomsday scenario.
Key takeaways:
- The term p(doom) is a buzzword that stands for "probability of doom," referring to the odds that artificial intelligence will cause a doomsday scenario. It is used as a common scale to describe where people stand on the question of whether AI is going to be harmful to humankind.
- The scale runs from zero to 100, with a higher score indicating a stronger belief that AI is not only capable of eliminating humankind, but is likely to succeed in doing so.
- Many people active in the AI field have volunteered a p(doom) score, even if they believe it's a speculative question. A recent survey of AI engineers found that their average p(doom) score was 40.
- Despite the widespread use of p(doom) scores, it is acknowledged that the actual probability of AI causing harm is unknown and can be complicated by various factors, such as the definition of 'doom' and the method of AI's potential takeover.