Demis Hassavis thinks his dog is conscious. Today's AI, he says, does not.
The Google DeepMind co-founder says that today's AI is very smart, but he doesn't feel like it's "at all conscious yet." His dog is upside down:
“My pet dog - he's adorable, and he's quite aware, I think. He dreams and all, but he's not that smart. ".
So intelligence and consciousness can exist separately, he told Reed Hoffman at the Scale Masters Summit in October. He wants labs to build smart tools, not conscious creatures. An AI with a will of its own could refuse when people tell it to stop or turn itself off, he claims.
He will first use artificial intelligence to study the brain and discover what consciousness is, which he thinks could take five or 10 years. The company is still not ready for the other questions, he says: "Should we worry about what we do with them in this box? Are they slaves of some kind? ".