I can see how the millions of games played by AlphaGo during practice can generate certain knowledge that’s useful. But we actually could learn more about our games. I looked at its games, and I wrote about them in an article that mentioned chess as the “drosophila of reasoning.” Every computer player is now too strong for humans. Returning to chess, what do you make of AlphaZero’s style of play?
And unfortunately we have enough political problems, both inside and outside the free world, that could be made much worse by the wrong use of AI. Yes, it's a unique tool because it can augment our minds, but it's a tool.
It's not a harbinger of utopia or dystopia. We have to actually look and just understand how we can fix it, not say “Oh, we can create AI that will be better than us.” We are somehow stuck between two extremes. I think it added too much advantage to bishop in terms of numbers."ĪI is like a mirror, it amplifies both good and bad. So machines will be dominant in the closed systems, whether it's games, or any other world designed by humans.ĪlphaZero "values bishop over knight. I don't believe that machines are capable of transferring knowledge from one open-ended system to another. And I think AI is just a great tool to achieve something that was impossible 10, 20 years ago.
At the same time, what's the difference? We have always invented machines that help us to augment different qualities. We are comfortable with machines making us faster and stronger, but smarter? It’s some sort of human fear. What we understand today is AI is still a tool. Even the best computer experts, the people on the cutting edge of computer science, they still have doubts about exactly what we're doing. We don't know exactly what intelligence is. How much progress do you think we’ve made toward human-level AI? You put the right machine in the right space to do the right task. Just basically push them in one direction or another, and they will do the rest of the job. You just have to nudge the flock of intelligent algorithms. With AlphaZero and future machines, I describe the human role as being shepherds. And that's the future of human-machine collaboration. Because the gun is so powerful, a tiny shift can actually make a big difference. Now a 1-millimeter change in the direction could end up with a 10-meter difference a mile away. Imagine you have a very powerful gun, a rifle that can shoot a target 1 mile from where you are. "Technology is the main reason why so many of us are still alive to complain about technology." So what you should do, you should try to get your engine to a position where AlphaZero will make inevitable mistakes. So I think it added too much advantage to bishop in terms of numbers. It sees over 60 million games that statistically, you know, the bishop was dominant in many more games. For example, it values bishop over knight. And I believe it has made some inaccurate evaluations, which is natural. People ask me, “What can you do to assist another chess engine against AlphaZero?” I can look at AlphaZero’s games and understand the potential weaknesses. But a big star in medicine will like to challenge the machines, and that destroys the communication.
A person with decent knowledge will understand that he or she must add only a little bit. If you have a powerful AI system, I’d rather have an experienced nurse than a top-notch professor. At the end of the day it's about combination. There are different machines, and it is the role of a human and understand exactly what this machine will need to do its best.