Bill Gates: The Three Phases Of The Robot Revolution

the robot revolutionWe can all agree to disagree with much of the discussion about robots displacing our jobs in the very near future. It is happening, that’s a fact. What’s hard to predict is when and how fast it will happen.

In previous episodes of my podcast (here and here) we talked about the rise of the machines and what it might mean for society. The way I think about the evolution of artificial intelligence is that it will evolve in phases: easy and complex.

The uses cases we see and read about today have to do with easy tasks that are completed via pre-determined commands and pattern recognition. But we still have ways to go before an AI takes on more complex tasks that deal with fears and desires; like in the movie Her.

On a recent interview, Ezra Klein asked Bill Gates about the state of artificial intelligence. Bill Gates provided a good framework to think about the evolution of artificial intelligence:

EK: I know you take the risk of creating artificial intelligence that ultimately turns against us pretty seriously; I’m curious where you think we are in terms of building artificial intelligence.

I know there’s a lot of disagreement in the field about, are we 40 years away? Are we 500 years away? What do you think is the state of AI research right now, and when do you think it will really begin feeding back into the economy and into innovation?

BG: Well, with robotics, you have to think of three different milestones.

One is just pure labor substitution for jobs that are largely physical and visual manipulation — driving, security guard, warehouse work, waiter, maid. That threshold — I don’t think you’d get much disagreement that over the next 15 years the robotic equivalents in terms of cost, in terms of reliability, will become a substitute to those activities. So that’s the first stage, and you’d get less variance in that prediction.

Then there’s what we think of as “intelligent activities” — things like writing contracts, or doing diagnoses, or writing software code. When will the computer start to infringe? “Infringe” is a pejorative word. When will it start to have the capacity to work in those areas? Some might say 30 years — I might be there. Some might say 60 years. Some might not even say that.

Then you have the sort of third threshold, where the intelligence involved is dramatically better than humanity as a whole, and there you’re going to get a huge range of views, including those who think it won’t ever happen. You have Ray Kurzweil, who says it will happen, I think, at midnight on July 13, 2045, or something like that, and it will all be good. Then you have people who say it can never happen. Then you have the group that I’m more among, that says, “Okay, we’re not able to predict it, but it’s something that people should start thinking about.” We shouldn’t restrict activities or slow things down, but the potential that that exists — even in a 50-year time frame — means it’s something to be taken seriously.

But those are different thresholds, and the responses are different.

Bill Gates had some great thoughts about climate change, energy and innovation as well; I recommend you listen to the whole interview.