The Coming Of The Bots

It was in 1948 that Cliff Gladwin, an English bowler and tail ender, uttered the famous phrase Cometh the Hour, Cometh the Man. A self-congratulatory statement at the end of a cliffhanger of a cricket match. England had just beaten South Africa in the very last ball of the Durban Test, courtesy a streaky leg-bye that came off Gladwin's pads. No doubt Cliff Gladwin was mighty pleased with his winning contribution. His rather generous self-assessment, in hindsight, became a great quotable quote.

Fast forward to fifty years later, 1997. I was at my first job as a young research scientist at the IBM Thomas J. Watson Research Center in New York when Deep Blue, IBM's chess playing machine, stunned the reigning world chess champion Gary Kasparov, 3.5 - 2.5. The day was May 11, and basking in the reflected glory of fellow computer science warriors I remember telling myself, Cometh the Hour, Cometh the Bot. The age of AI (Artificial Intelligence) had dawned and begun to lay waste to centuries of carefully-constructed human vanity. Never again would anyone cite chess grandmastery as clinching evidence of human supremacy over the machine.

Jump nineteen years to 2016, and the AI genie is well and truly out of the bottle. Magnus Carlsen, the current human chess champion, stands overtaken in rankings by at least 50 chess players made of silicon. Driver-less cars are causing fewer accidents than human-driven ones, leading to lower insurance premiums, and raising fundamental questions about the future of the auto-insurance industry. The Henn-na hotel in Japan has opened its doors with humanoid robots as staff, promising its clientele the ultimate experience in efficiency. DuoLingo is disrupting the market for paid language teachers by dispensing free and unlimited foreign language learning on its elegant app. From playing advanced games to driving cars to serving guests to teaching languages, machines are doing what humans have done for centuries, only better.

Does the future of human job security lie in being able to conceive, design and develop intelligent machines? After all, bots are created by humans, right? Not true. Given the right conditions, bots can even teach themselves to become extremely intelligent. Just last month, Lee Sedol, world champion in Go, a far more complex game than chess, was beaten by a computer program called AlphaGo. Sounds just like Deep Blue beating Kasparov? No big deal? Well, the truth is stranger than fiction. AlphaGo was not programmed by a team of elite human programmers the way Deep Blue was. Instead, as an AI program it used deep machine learning and reinforcement learning techniques to master the game of Go by continually playing millions of boards against itself, learning from its own mistakes, correcting and upgrading itself at every step, and eventually figuring out how to best a human world champion like Sedol at his own game. It simply taught itself to win.

The future of the global workforce, given this backdrop, looks dicey. For a chillingly clinical preview of the coming chasm between employees and employers, look no further than Humans Need Not Apply, an eye-popping YouTube video by C.G.P. Grey. Here is the long and short of it. According to Grey, while hundreds of millions of global youth are going to college today in the hope of landing lucrative jobs tomorrow, tens of thousands of employers are doubling up on AI and automation to substitute error-prone expensive humans in their processes with error-free inexpensive machines. Most economic growth from this point on would be high on automation and low on job creation. Take that!

And that may just be the tip of the proverbial iceberg. AI expert and roboticist David Levy forecasts that by 2050 love and marriage with robots will be as normal as it is with other humans. Society will progress to the point where bots will become one more gender. In such a world, AI may be the ultimate answer to human deficits in the areas of intimacy, relationships, and companionship. And unlike humans who are fallible and buggy, programmable bots could give us a better shot at programming for perfection, a priori.

There is a scene in The Imitation Game, the 2014 film on the life and times of Alan Turing. The police detective interrogating the father of AI asks him the mother of all loaded questions: So tell me Professor, can machines really think? The great man's reply is a variation on his famous Turing Test. It really does not matter whether a machine can think. What matters is whether a machine can imitate a human so exceedingly well that one can no longer tell whether it is a human or a machine. Ask Kasparov. After losing to Deep Blue, he said: Deep Blue sees so deeply, it plays like God.