Who: Stuart Russell is a Distinguished Professor of Computer Science at the University of California at Berkeley, holder of the Smith-Zadeh Chair in Engineering, and Director of the Center for Human-Compatible AI and the Kavli Center for Ethics, Science, and the Public. Russell, a speaker at the Tallinn Digital Summit Nov. 5-6, is the president and co-founder of the International Association for Safe and Ethical AI (IASEAI), a global membership-based nonprofit that aims to promote research, shape policy, and ensure AI systems operate safely and ethically. the problem of keeping increasingly capable AI systems under human control.
Topic: The problem of keeping increasingly capable AI systems under human control.
Quote: “I would like to disabuse people of the notion that open means it is safe. AI companies go to some lengths to train these things not to misbehave; that is what the post-training phase is about – teaching AI to be polite and helpful, and then adding external guardrails. With open weights, it is trivial to remove the guardrails and undo the post-training that ensures that AI does not act like a feral psychopath. If anything, it needs more controls. What happens when 5 billion people have open access to technology that can bring down the infrastructure of a country and can do anything they want with it?”
To read the full article and gain access to over 1000 stories about global innovations become a paying subscriber to The Innovator. Click here to sign up.
