Google's AlphaGo Defeats China's Go - What does this imply about our future?
Google DeepMind: Ground-breaking AlphaGo masters the game of Go [Source: https://www.youtube.com/watch?v=SUbqykXVx0A]
Many Westerners hold the belief that chess is the most complex game in the world. They don't know much about 'Go', an ancient Chinese board game. To even play it (not to speak of winning), an individual needs an incredibly complex brain that until now most scientists believed only human beings could possess.
But the world of Artificial Intelligence just got a tremendous boost a week ago when scientists at the Google DeepMind laboratory in London announced that it has created a computer armed with the most advanced Artificial Intelligence that beat the human European Go champion 5 games to 0. The results have just been published in the article 'Mastering the Game of Go with Deep Neural Networks and Tree Search' that appeared in the January 27, 2016 issue of Nature, the international weekly journal of science.
I know what some of you might be thinking: Deep Blue, the supercomputer created by IBM, defeated the human chess champion Gary Kasparov nearly 20 years ago, so this is old news. But far from it. Google DeepMind's creation, AlphaGo, is a much different machine and has taken the field of A.I. about a decade ahead of where experts thought it would be in 2016.
Each time Deep Blue was ready to move a piece on the chess board, it would use what's commonly known as a look-ahead search function that would calculate within seconds every single possible outcome of the game by moving any one piece. It's mind-boggling to think that Deep Blue could accomplish this feat every time it was its turn to make a move.
AlphaGo supersedes this. The game of Go is considered to be even more challenging than chess in terms of the sheer number of moves and board positions the player must evaluate. AlphaGo uses a collection of value networks that evaluate board positions as well as policy networks that select the moves. Scientists from DeepMind say that these networks are "neural," meaning they can be trained by learning to play a number of games that are supervised by humans and then allowed to play by itself, using what it learned in previous games. The big news is that AlphaGo does not use the look-ahead search function in its mastering of Go. Instead, it utilizes a new search algorithm that combines Monte Carlo game simulation with the A.I.-style value and policy networks.
2025 - This is the year, according to most experts, when machines with A.I. will be able to have independent thoughts, actions, and conversations with humans without any human supervision. With A.I. milestones such as AlphaGo, we may get there even sooner. Advances in A.I. is showing that A.I. can make "big picture" and take philosophical decisions. In the meantime, the debate rages on about whether A.I. will complement and benefit humanity in the future, or create a dystopia of sharp inequalities vis-à-vis wealth, knowledge, opportunities, skills, and impact on nature.
Frankly, I couldn't be more optimistic, because as A.I. takes over old, repetitive jobs, new ones that demand greater human skills will be created in the wake of this impending market disruption. Another equally important aspect is being able to balance nature's cycle of maintaining equilibrium, which will need human knowledge and creativity to help remodel skills to sustain nature holistically. True, it is not going to be an easy transition into a world that will be irrevocably digitized, probably even transformed by A.I. capabilities. We will first need to develop 'coping mechanisms' to adapt to this new environment, and upskill to new kinds of jobs gradually. Governments across the world will need to intervene with regulations and schemes to protect some jobs and skills, and even contain unbridled advancement in A.I., if any private organization pursues it without considering the benefit of larger society.
It is not a question of utopia or dystopia (that would be oversimplification), but an impending reality that is exciting and needs to be shaped by our collective conscience and future needs. At Infosys, we believe that the world is on the cusp of a human revolution, fuelled by unprecedented technological advancement, including A.I. This is also what drives our corporate philosophy ─ that technology will amplify human potential, enabling us to realize our creativity and unfetter our imagination, so we can be much more than what we believe we are capable of, and this will be made possible by continuous investment in education and lifelong learning.