AI pioneer Gary Marcus has been a voice of caution and reason against AI's apocalyptic doomsayers, but is eager to point out artificial intelligence's real dangers.

"When you talk about like extinguishing the whole species, that's actually pretty hard," said Marcus, as he discussed the industry's latest developments and the risks they imply for the public, democracy, and world geopolitics.

However, he told AFP that "we have a lot of work to do, I think, before we can feel confident that we've got this under control. We don't right now."