Steven Strogatz (NYTimes) on a future for AI via AlphaZero

AlphaZero is now the undisputed champion of Go and now of chess. It recently battled Stockfish, the former chess computer heavyweight, and in that series of 100 matches, it won 28, drew 72, AND LOST NONE.

Lets hear that again. AlphaZero, the deep learning computer originally designed to play and beat human players at Go, the ancient board game, has recently been redesigned in a couple ways: 1) to take the original game rules AND NO HUMAN EXPERIENCE as its starting point, and 2) now can receive the rules for almost ANY game (in this example, chess) as its starting point. Then the programmers set AlphaZero to play itself AND LEARN THE STRATEGIES of the game by brute force and whether each strategy led to a victory or defeat. 

AlphaZero, having spent time playing itself millions of times and having discerned and taught itself the principles of chess, it only considered 60,000 moves per second instead of 60 million by Stockfish. It played smarter and faster.

“AlphaZero had the finesse of a virtuoso and the power of a machine.”

But, can it teach us its insights? No. Perhaps the most troubling paragraphs in this article is:

“What is frustrating about machine learning, however, is that the algorithms can’t articulate what they’re thinking. We don’t know why they work, so we don’t know if they can be trusted. AlphaZero gives every appearance of having discovered some important principles about chess, but it can’t share that understanding with us. Not yet, at least. As human beings, we want more than answers. We want insight. This is going to be a source of tension in our interactions with computers from now on.”

I am both heartened and disturbed by this. Heartened in that AI is on the launch pad to apply itself to all kinds of human challenges that have been difficult to solve until now. Disturbed also; how long will AlphaZero and its contemporaries need human insight and input before it’s always-accelerating capability outstrips our brains’ hardware and our ability to keep up and be relevant?

CMIO’s take? I have no take. I’m gonna wait for my auto-correct from Siri to get smart enough to finish writing this post.

Author: CT Lin

CMIO, UCHealth (Colorado); Professor, University of Colorado School of Medicine

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s