• Misalignment is the source of the concern that people have had about AI.
• Going right back to Alan Turing, who was the founder of Computer Science in a 1951 lecture. He said once the machine thinking method had started. Thinking. It would leave our feeble powers far behind. And we should have to expect the machines to take control. So they take control not because they’re evil or because they spontaneously develop consciousness or anything like that. It’s just because we give them some objectives that are not aligned with what we want the future to be like. And because they’re more capable than us, they achieve their objectives and we don’t, right? So we set up a chess match which we proceed to lose.
• So in order to fix that problem, I’ve been following a different approach to AI, which says that the AI system, while it’s only objective, is to further the interests of human beings, doesn’t know what those are and knows that it doesn’t know what those are. It’s explicitly uncertain about human objectives.
• And so to the extent that there’s a moral theory, it’s simply that that the job of a AI system is to further human interest. It knows that it doesn’t know what those are, but it can learn more by conversing with us, by observing the choices that we make and the choices that we regret, the things we do, the things we don’t do.
• So this helps it to understand what we want the future to be like. And then as it starts to learn, it can start to be more helpful.
• There are still some difficult moral questions mean. The most obvious one is the it’s not one person’s interest. It’s not one set of values. There’s 8 billion of us, so there’s 8 billion different preferences about the future and how do you trade those off?
• And this is a two and a half thousand year old question, at least, and there are several different schools of thought on that.
• And we better figure out which is the right one, because we’re going to be implementing it fairly soon.
• And then there are even more difficult questions like, well, what about not the 8 billion people who are alive, but what about all the people who have yet to live? How do we take into account their interests? Right, right. What if we take actions that change? Who’s going to live? You change the number of people who are going to live.
• For example, the Chinese policy of one child per family probably eliminated 500 million people already. Now they never existed. So we don’t know what they would have wanted, but how, you know, how should we make that type of decision?
• Right. These are really difficult questions that philosophers really struggle with.
• But when we have AI systems that are sufficiently powerful that they could make those decisions, we need to have an answer ready so that we don’t get it wrong.

• And so, you know, part of the job of of the cavalry center that you mentioned at the beginning is to bring philosophers, social scientists, political theorist, legal theorists and A.I. researchers and gene editors and neurotechnology people together to start figuring out answers to these questions before it’s too late.
• Because, you know, we are going to have gene editing. Do we want to allow people to pay to have their children become more intelligent than they would otherwise have been? Do we want neurotechnology that allows us to connect to minds together and turn them into a single conscious entity?
• Well, we better figure it out because otherwise the market is going to make that decision.

• So in terms of the existential risk, which would come from, you know, as Alan Turing said, the machines taking control because once they take control, so to speak, there’s really no longer anything the human race could do to ensure its continued survival.
• It might be that the machines allow us to continue or not, right?
• We will be in the same position as the gorillas are with respect to humans, right. There was this little thing that happened a few million years ago where one branch of the primates ended up more intelligent than the others. And so all the other branches now continue to exist, basically because we allow and some of them have already gone extinct as a result of competition with humans. So we don’t want to be in that situation.
• I believe it’s possible to develop AI systems that are provably safe and beneficial, that we can retain control over that actually want to be switched off. That’s a really important thing, right? If we want to switch it off, it needs to want to be switched off. And that’s a consequence of the theory that I’m working on. But it’s not a property of the kinds of systems that we’re building now.
• So on the other questions, you know, what is the future of our coexistence with machines? What types of lives will people have? How will they continue to have valuable economic roles? When I can do pretty much all the jobs that humans can do, I think that’s a really important question for policymakers, because my guess is that the value that we can provide will be much more of an interpersonal nature, but it’s not going to be the value that a factory worker can provide because as we know, those types of jobs are already being automated out of existence. It’s not going to be in routine clerical work.
• I mean, a simple way of putting it. I know, Jerry, you don’t necessarily agree with with this line of argument, but it shows if you can no longer sell physical labor and you can no longer sell mental labor, it’s not clear that there’s another thing, right, that that human race can fall back on, except we might call it interpersonal or emotional or empathic capabilities, where we have this sort of intrinsic comparative advantage over machines because because we know what it’s like.
• Well, right. And I give this example in the book. Right. What’s it like to hit your thumb with a hammer? Right. Who’s done that? Right. Most of you. And someone who hasn’t done that. Right? A few. Okay. Well How would you find out what it’s like if you didn’t know or you would just hit the phone with a hammer? You say, Oh, now I get it. Now I understand why people are so upset when they do that, right.
• But there’s nothing a machine can do to find out what it’s like, right? They can at best make empirical correlations and assume that it’s unpleasant, but they don’t know what it’s like. They don’t know what it’s like to be left by your lover. They don’t know. It’s like to lose a parent or to lose a child or to lose a job or to be promoted or any of the feelings of what it’s like to be human. And so there we have this comparative advantage, and there are also things that we just don’t want to be done by machines right?
• I imagine that at some point in the future there’ll be a profession that we might call lunch to someone who’s really, really good at having lunch with you. Right where you have lunch with them, you go away feeling much better about yourself. Entertained, amused, wiser, more positive, and so on. Right. And you won’t get those feelings if that, if that was a robot. Well so we’ll see.
• The difficulty is that most of these interpersonal jobs right now are low status because they are they are not based on real scientific understanding. If you compare babysitting with orthopedic surgery. Right.
• My children are actually more and more important to me than my arms and legs. Right. But we pay the orthopedic surgeon 100 times or 1000 times as much per hour as as the babysitter.

OpenAI’s question-and-answer chatbot ChatGPT has shaken up Silicon Valley and is already disrupting a wide range of fields and industries, including education. But the potential risks of this new era of artificial intelligence go far beyond students cheating on their term papers. Even OpenAI’s founder warns that “the question of whose values we align these systems to will be one of the most important debates society ever has.” How will artificial intelligence impact your job and life? And is society ready? We talk with UC Berkeley computer science professor and A.I. expert Stuart Russell about those questions and more. Photo courtesy the speaker. April 3, 2023 Speakers Stuart Russell Professor of Computer Science, Director of the Kavli Center for Ethics, Science, and the Public, and Director of the Center for Human-Compatible AI, University of California, Berkeley; Author, Human Compatible: Artificial Intelligence and the Problem of Control Jerry Kaplan Adjunct Lecturer in Computer Science, Stanford University—Moderator 👉Join our Email List! https://www.commonwealthclub.org/email 🎉 BECOME a MEMBER: https://www.commonwealthclub.org/memb… The Commonwealth Club of California is the nation’s oldest and largest public affairs forum 📣, bringing together its 20,000 members for more than 500 annual events on topics ranging across politics, culture, society and the economy. Founded in 1903 in San Francisco California 🌉, The Commonwealth Club has played host to a diverse and distinctive array of speakers, from Teddy Roosevelt in 1911 to Anthony Fauci in 2020. In addition to the videos🎥 shared here, the Club reaches millions of listeners through its podcast🎙 and weekly national radio program📻.