Facebook. Google. Apple. Baidu. Not a day goes by when we don’t hear of something related to artificial intelligence in the news. But AI (sometimes confused with machine learning, which is simply a technique within AI) wouldn’t be where it is today if it weren’t for one seminal event in 2016: AlphaGo beating Lee Sedol.
It’s the tech equivalent of Leicester City winning the Premier League, or the Golden State Warriors winning 73 games in the NBA. It’s the unimaginable happening, right in front of our own eyes.
In March last year, an AI program that trained itself to play the ancient game of Go beat the 18-time world champion. The reason it was such a feat for AI, was because Go is about feel, strategic judgment and winning multiple battles across the board – and a computer cannot simply memorise all possible combinations of board pieces, assess the situation, construct and execute a strategy to win, like chess.
So the programmers of AlphaGo, from Google DeepMind, set up the basic heuristics of the game, allowed AlphaGo to analyse previous games and then split its brain so it could play itself millions of times.
Imagine the 1983 film War Games in which a computer was taught to learn by playing tic-tac-toe against itself. This time, however, the game is Go.
The results were stunning. AlphaGo beat Lee Sedol convincingly, 4 games to 1. So much for human intuition. And to add insult to injury, the Korea Baduk Association awarded AlphaGo the highest Go grandmaster rank.
This event was important because it made clear that AI was here to stay. Of course, AI is also receiving a big push from Facebook and Google as they invest billions of dollars through OpenAI and DeepMind to democratise the technology – not to mention countless acquisitions of AI talent. But the corporate activity merely proves that big-business has plans for AI, whereas AlphaGo beating Lee Sedol signals the possibility that AI could indeed be more capable than humans.
And that is a conceit that an AI investor like me is comfortable with. There are certainly some things that machines are better at than humans. In fact, there are many things that machines should be better at.
Self-driving cars is one example. Imagine a world where traffic accidents are a relic of the past and very few lives are lost. Imagine a future where your cancer prognosis is not misdiagnosed by multiple doctors and visitations, but instead you are told in one go what cancer you have, how long you have to live and what stage it is at. Imagine a future where your frustrated customer service hotline calls are replaced by an AI that recognises everything you say and responds in the best way. In many ways, the human variability that makes us all frustrated will be replaced with machine predictability.
WATCH: Google unveils prototype of self-driving car
Many people are aghast when I mention the predictability versus variability argument. “Human fallibility is great!” they say. “It’s what makes us alive!” Bollocks. Human fallibility and variability are great for art, where it is a necessary requirement to produce even greater art. It’s great for startups, when we are looking for the next big innovation and we don’t know where to look. But it’s absolutely terrible for things like cancer detection, driving, aviation and a whole host of things that don’t need human intervention or decision-making.
I am biased, of course, but I’ll make two predictions about AI for 2017 and beyond. One, AI will be the biggest thing since sliced bread.
You can consider corporate activity like Facebook, Google, Snapchat, Baidu as a signal of trust. You can also trust the direction of car companies like GM, Mercedes-Benz, Uber, and Tesla towards autonomous cars. And you can trust academics, the White House and other longitudinal thinkers when they say past academic research, computational power and large accessible data sets will together create a perfect storm for AI impacting our lives. All signals show AI will be big.
Another prediction: Asia will be big in AI. If you look at the data trends, there is a tonne of AI research done in China, Japan, South Korea and Singapore. In fact there are more cited publications in China than in the United States. The main problem, however, is the cultural barriers to sharing this knowledge between Asia and the rest of the world, many of which stem from language differences.
But there is another barrier. In the West there is a tenancy, even incentives, to commercialise research, but the same dynamics are not present in Asia.
Nevertheless, the underlying core principle can not be denied – AI’s potential in Asia is rising and Hong Kong would do well to find a way to ride this wave. ■
Tak Lo is the founder of Zeroth, a Hong Kong-based accelerator programme for Asia-focused AI and machine learning startups