If it’s a self-driving car, the passenger puts in the objective: I want to be at the airport. So if it’s a chess program, you give it the goal of beating your opponent, of winning the game.
To answer that question, we have to understand: how are AI systems designed? What do they do? And in the Standard Model you build machinery, algorithms, and so on that are designed to achieve specific objectives that you put into the program. What’s the case that advanced AI could be dangerous for humanity? Stuart Russell Here’s a transcript of our conversation, edited for length and clarity. I met Russell at UC Berkeley, where he heads the Center for Human-Compatible AI, to talk about his book and about the risks posed by advanced artificial intelligence. And that means there’s a disaster on the horizon. AI systems, Russell argues in Human Compatible, care about only whatever we’ve put in as their objective. Humans care about a lot of things: fairness, law, democratic input, our safety and flourishing, our freedom. Or a health care cost-saving system that discriminates against black patients because it anticipates that they’re less likely to seek the health care they need. Imagine a self-driving car with an “objective” to get from Point A to Point B but unaware that we also care about the survival of the passengers and of pedestrians along the way. If they hit on a strategy that fits that objective, they will run with it, without explicit human instruction to do so.īut with this approach, we’ve set ourselves up for failure because the “objective” we’ve given the AI system is not the only thing we care about. AI systems, he notes, are evaluated by how good they are at achieving their objective: winning video games, writing humanlike text, solving puzzles. In a new book, Human Compatible, he explains how.
He has also, for the last several years, been warning that his field has the potential to go catastrophically wrong.
Stuart Russell is a leading AI researcher who literally wrote (well, co-authored) the top textbook on the topic.
0 kommentar(er)
