Not So Fast! Why Robots May Not Be Coming

Robots were supposed to take over the universe. Singularity - the end of human history - was due in about a decades time. And, then, it was to be a different world altogether - one with self-knowing, self-replicating machines.

But it seems that future may have been postponed. 

If there is anything that the history of technological progress should teach us, it should be that we ought not to believe in the excited predictions that technology industry loves to make. It is part of their style - all those world-changing predictions! There is a vast difference between the rhetoric of Microsoft's seamlessly connected world and the reality of clunkiness of Windows; E-Commerce did well but the prognosis far outstripped the performance, and these are only the good examples. No aircraft ever fell out of the sky despite all the warnings of the Y2K doomsday: It did create a multi-billion dollar outsourcing industry though. And, there are those who are still waiting for the day when everyone learns online, or Powerpoint becomes engaging. 

Yes, technologists talk it up. And, the recent history of technology - a graveyard of great expectations - shows that, beneath such excitement, is not naive optimism, but calculating strategy. It's the Donald Trump principle: If you keep telling the same thing, it becomes true - until it isn't. Conjuring up Y2K was one of those great rope-tricks: A big industry was created without anyone getting hurt! So was Dotcom, and all the private Internet companies that came after. There is money to be made, a lot of money, in making predictions: The astrologers knew this for centuries, and the Futurists have taken over the realm.

Underlying all the assumptions about Robots taking over the world, there is a simple, rather unassuming, and famous Law - Gordon Moore's predictions about semiconductor capacity and cost ratio. Projected as one of the iron laws of the universe, this informs all the sophisticated looking charts and revered sounding papers churned out to predict technological future. Moore's Law was relative though: One may say it uncannily held for several generations, but that would gloss over the fact that it was revised. Revised for better, indeed: Gordon Moore predicted semi-conductor capacity doubling, at given cost, every two years; but since the eighties, it was doubling every 18 months!

However, the iron laws of the universe do not change, even for better. And, indeed, it is not holding up now - semi-conductor progress has slowed. Yet, our predictions about machine-learning are really the more optimistic scenarios based on Moore's Law, and its stepped down version (say, going back to double in two years rather than 18 months) would create a wildly divergent scenario. No one talks about it - there are billions of dollars invested in the rosy scenarios and it's just too big a boat to rock - and one optimistic prediction leads to another. But, even a basic reckoning of Moore's Law may tell us that Robots are not taking over themselves anytime soon.

This, of course, does not mean that it wouldn't hurt in the meantime. Under the cloak of inevitability, we have invested more money in building technological capabilities than human ones. Schools have suffered while start-ups came and gone vanishing billions of dollars with them. Therefore, indeed, machines would take over some jobs as the engaged and capable human workers couldn't be found. But even that is facing a backlash: One can't any longer reconcile popular democracy with the logic of reconfigured workplaces. We are at a crossroad: We either disable popular democracy; or democracy - or as we now call it, populism - would stop the incessant privileging of technology over human. And, either way, this means our date with Robots have to be postponed.