ARCHIVE: The Dream of AI Is Alive in AlphaGo

**Frank Chen** (0:00)
Look, we've had a lot of starts with artificial intelligence, and the vector has always been, hey, look, now that we've run this very sophisticated board game, now we're on the verge to do general purpose intelligence.
Look at the very sophisticated set of techniques they use to win this game. There's deep learning, and then there's decision trees, and then there's supervised learning. Look at all of the techniques. And maybe now, this time, it really is the dawn of the generalized intelligence.

**Sonal Chokshi** (0:30)
Welcome to the A16Z AI Podcast, where, normally, you'd be hearing an interview with a leading founder, engineer, or other voice in AI. But because it's a holiday week here in the United States, we're giving you a gift from the vast A16Z Podcast Archive. If you've been following AI for a while, you're probably familiar with AlphaGo, the system first released by Google DeepMind in 2015 and designed to master the Chinese board game Go and if you've been listening to the main A16Z Podcast for a while, you might have already heard this episode from back in 2016 when AlphaGo achieved a computer first by besting a high level champion in a five game match without a handicap. At the time, it was a very big deal. I know I had never heard of Go master Lee Sedol before this and eight years later, his name is still etched in my brain. But we're replaying this episode as more than a mere trip down memory lane.
Rather, as we regularly see new game changing capabilities from generative AI models that make us question the line between human and machine intelligence, it's good to have some reminders of earlier state of the art systems. What made Go an appealing challenge a decade ago was that, unlike, say, chess, which specialized computers had long since mastered, Go was remarkably complex and rewards creativity. Previous types of expert systems couldn't master the game. As DeepMind explained in 2018 when it retired AlphaGo, the game is, quote, a Google times more complex than chess with an astonishing 10 to the power of 170 possible board configurations. That's more than the number of atoms in the known universe, unquote. Even Lee Sedal acknowledged that, although he initially thought AlphaGo was just great at calculating probabilities, its gameplay convinced him the system really was creative. So, with that background, enjoy listening to a16z's Sonal Chokshi and Frank Chen and board member Steven Sinofsky discuss why this was such a big deal.
And note that the transformer architecture, which underpins so many of today's foundation models, first hit the scene in a paper about a year later.
As a reminder, please note that the content here is for informational purposes only, should not be taken as legal, business, tax or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a16z fund. For more details, please see a16z.com/disclosures.

**Sonal Chokshi** (2:57)
Hi, everyone. Welcome to the a16z podcast. I'm Sonal. And today we have two partners from Anderson Horowitz. We were just having an informal conversation in the hallway, literally, around machine learning and AI. I'm Steven Sinofsky, a board partner for a16z, gave a presentation on the evolution of machine learning, and Frank Chen recently put out a tweet storm on why Google's DeepMind algorithm beating least at all is so significant. And they were sort of talking about like, oh my God, we've been here before. But it's not going to be all backward looking, because I think the point is that the evolution is what's... Why now?

**Steven Sinofsky** (3:30)
Right, and also, you know me, I love to put things in context, because there's always lessons to be learned and patterns to avoid and not avoid, and patterns, that's a keyword for today, too.

**Sonal Chokshi** (3:38)
Okay, well, let's start talking about those patterns.

**Frank Chen** (3:40)
Well, maybe let's start with the big go victory, which is it got people really fired up. It dominated the press for a little while, and you might be wondering, what is the big deal? Computer program won another board game. A board game, like what could be less relevant to everyday life, right? And we've seen this before. We started with Tic-Tac-Toe, and we got to checkers, and we got to chess, and then Watson even won Jeopardy, and now here we are talking about another board game. So, like, it's kind of irrelevant to everyday life, isn't it? And the surprising thing is, look, we've had a lot of starts with artificial intelligence, and the vector has always been, hey, look, now that we've run this very sophisticated board game, chess, checkers, whatever, now we're on the verge to do general-purpose intelligence.