Lessons from the Lee Sedol vs. AlphaGo Match
On Sunday, World Go champion Lee Sedol beat Google’s AlphaGo against all odds after suffering three consecutive defeats to the supercomputer. Anyhow, the Lee Sedol vs. AlphaGo match has opened a new chapter in the history of artificial intelligence (AI). It is all that matters. Just like calculators made abacuses a relic of the past, AI outsmarting world Go champions was bound to happen someday. Still, AlphaGo has exhibited its weaknesses.
Though some say that if China's No.1 Go player Ke Jie played against AlphaGo, things would turn out differently, I beg to differ.
In the first three games of the historic five-game challenge match between humanity and AI, AlphaGo has proven that it can outwit one of the world's best players of the ancient Chinese game of Go in terms of the ability to learn and adapt. .
The first three games have taught us that AlphaGo does not play Go the way human players normally do. AlphaGo is aggressive and thinks outside the box to ambush and blindside the opponent.
Therefore, first of all, the golden rules of playing Go should be reestablished. Second, collective intelligence should not be underestimated. Third, the possibility of winning should be taken into account. Fourth, South Korea should spur on its efforts to develop AI.
The previous golden rules of playing Go have turned out to be an outgrowth of short-sighted gambits which worked well within certain boundaries. In other words, they were based on “greedy algorithms.” The historic Lee Sedol vs. AlphaGo match has shown that what seems to be a “bad” move at a glance could be a well-thought-out, far-sighted “good” move. It indicates that Go masters have thus far lived like big fish in a little pond, pursuing “local optimum” rather than “global optimum.”
Collective intelligence of a group of brainy people has played a pivotal role in the development of AI. When smart people come up with various solutions, classification algorithm “boosting,” one of the props of AI, kicks in to find the best solution among them. It is like you are given an opportunity to consulting with 100 people before answering each question in a quiz show. Engineers at Google DeepMind say that AlphaGo factors in all the possibilities and makes moves that maximize its probability of winning. Playing a hunch will never result in scoring a victory against AlphaGo.
When IBM's Deep Blue and Watson pushed the envelope of humans, South Korea was cynical about AI; government official balked at the idea of channeling funds specifically into AI researches, saying that funding AI research was overlapping investments. We were busy picking low-hanging fruits for the immediate future, consequently neglecting to invest in basic science. To make robots do things at our bidding, we have to provide robots with good digital dictionaries, which we do not have.
Of course, AI is not all-powerful. If you think that AlphaGo has a high IQ, you are wrong. AlphaGo’s deep learning ability and super-fast calculating ability have defeated Lee Sedol three times in a row. Still, computer cannot describe what lions are doing when a photo of lions are shown. Computers cannot answer tricky math questions that Korean senior high school students are wrestling with at school. Nevertheless, AlphaGo won three games in a row before losing one on Sunday. It is dumbfounding.
Although AI is not complete, autonomous cars are in. Robo advisers which generate investment advice and manage your portfolio are replacing humans. AI is crucial to big data technology and fintech.
AlphaGo does not know the ultimate solution every time. However, it has proven that AI is a cut above than humans. As such, AI has a long way to go
Having witnessed Lee Sedol losing three games in a row, the South Korean government belatedly said that it would intensively promote the development of AI technologies, which is nothing but an act of ostrichism. It will take some time for the S. Korean government to behave like its counterparts in the developed world. Developed countries quietly but preemptively choose the right path forward and start their journey on the path.
(Prof. Kim Hyoungfirstname.lastname@example.org)