If we are to take seriously the threat dealing with humankind, regulatory authorities require the power to ‘remember’ released designs, along with evaluate leading, not delaying, indications of threat, composes Prof John McDermid
Re Geoffrey Hinton’s problems concerning the dangers of expert system (‘Godfather of AI’ shortens odds of the technology wiping out humanity over next 30 years, 27 December), I think these problems can best be reduced via joint study on AI safety and security, with a function for regulatory authorities at the table.
Presently, frontier AI is examined post-development making use of “red groups” that attempt their finest to generate a unfavorable end result. This method will certainly never ever suffice; AI requires to be developed for safety and security and analysis– something that can be done by making use of proficiency and experience in reputable safety-related markets.
发布者：Guardian Staff，转转请注明出处：https://robotalks.cn/collaborative-research-on-ai-safety-is-vital-letters-2/
