🔒 Make yourself and your family AI-scam proof, step by step → https://neuralnutshell.com
Geoffrey Hinton, a Nobel laureate and luminary of artificial intelligence who is thought to be one of the founders of modern deep learning asserts that today’s large language models are now developing their internal representation of the world instead of simply memorizing texts; reason more than a preschool aged child; and are continually moving toward superhuman capabilities. Hinton addresses in this video: how Asystems naturally pursue multiple subgoals in addition to their primary objective i.e., take control and avoid termination; the research that shows how goal oriented models exhibit deceptive behaviours; whether software based AI systems can become "immortal" when their model weights are duplicated; how equally capable AIs can have unprecedentedly rapid information sharing; and to what extent will subjective experience or consciousness be reserved for humans. Ultimately, he warns that society has yet to fully account for the ramifications of creating ever more sophisticated digital intelligence.
Δεν υπάρχουν σχόλια:
Δημοσίευση σχολίου