When AI Meets A Robot

The Singularity May Be Boring

The Singularity

The explosive growth in the power of AI — both in terms of breadth as well as depth — has people worried (or overjoyed — depends on how you look at it) that we will soon reach a stage where AI not only outstrips human intelligence in all areas, it becomes recursively able to improve itself, without further human intervention.

When that happens, it is claimed, growth in intelligence will become virtually infinite, because AI “neural” speeds are millions of times faster than that of a human brain — and the AI does not have to pause to eat, drink, or have sex.

When that happens, we will have hit a singularity. Analogous to a black hole compressing spacetime to an infinite point, AI could achieve more intellectual progress in a few hours, or minutes, than humanity has in millions of years.

AI and Robots

But wait, there is more…right now, we have robots carrying out specific tasks, while AI is excelling at more and more general tasks. What happens if you combine the two? You could conceivably have a situation where AI designs general purpose robots.

And somewhere at the end of that path, you can imagine a kind of nirvana on Earth, where the AI assigns a bit of its intellectual capacity to create robots that make life for humans easier.

Much, much easier.

As in…do everything that humans can do, only much better.

The End of History

No more plumbers, car mechanics, factory workers, farmers, lawyers, doctors, or babysitters — in fact, all human occupations would become redundant.

All human needs would now be taken care of by AI robots, so that we can all relax and watch Netflix the whole day, where even the programming will be all AI-generated. In fact, rather than selecting a movie at will, you would select a storyline, and the AI would generate a movie, with photorealistic rendered frames, all in real time. Heck, you could even change the storyline on the fly.

In such a world, we may ask, who will control all these robots? Will it be a few very, very rich people, who will control all the means of production, i.e. the AI robots — patents, factories, whatever?

Or will the AI robots become a force unto themselves, bypassing humans completely, and give us what we want, so that each one of us has an army of robots to serve us, including (yes, admit it, that is what you were thinking) comely / handsome sexbots?

This would be like the end of history as envisaged by Marx, except that it would be a supercharged version of communism — to each according to his needs, without the pesky requirement to first get from each according to his ability, because the robots will be doing all the producing and all the work for us.

But will AI really get us to a singularity?

The Devil is in the Details

Consider how AI is trained, at least for now. It is trained on the sum total of human knowledge and information available out there, from which it seeks patterns, and learns to reproduce these patterns in ways that seem credibly human, and credibly intelligent.

I may have simplified a lot, and no doubt an AI expert would quibble about the details, but that is essentially the case. So how will AI extend its “intellect” beyond that?

Imagine for a moment that the AI singularity had been reached in 1980, which means that we would now be living in Netflix heaven. You would be able to run many versions of a James Bond movie, with multiple iterations at every decision point.

But would you have Harry Potter?

Exponential Growth

Let’s look at it from another angle. Human intelligence, as represented by the technological forces that we have at our disposal, has already been progressing at an ever-increasing rate, i.e. exponentially.

Assuming a divergence from other apes around five million years ago, we have the “domestication” of fire around one million years ago, domestication of animals starting from fifty thousand years ago, agriculture at ten thousand years, writing, cities, and empires at five thousand, the industrial revolution at four hundred, and whatever is happening now measured over decades and soon, in years. Obviously, all figures are ballpark, but you get the idea.

In a sense, therefore, human growth is already exponential, like this:

And the people who see AI creating a singularity imagine that it will become like this:

But what if AI merely continues regurgitating what humanity has already achieved? Then, instead of a singularity, we will have this:

Sure, we may not have to work at all, and AI robots will be doing our every bidding…but unless the fundamental nature of AI changes, we will not have a singularity, we will have a plateau.

Instead of reaching for the stars, humanity will be reaching for the nearest sexbot.

And this is assuming that AI doesn’t kill us first.