Moravec’s paradox

Why do we (currently) live in a world where AI does art and intellectual labor while humans do hard physical work? I’ve often seen this sentiment or something similar echoed here and other places online. It’s essentially saying, more or less - “how do I live in a world in which Artificial Intelligence exists, but it gets to do the creative tasks while I toil away doing mean and demeaning physical labor 40+ hours a week”.

Most people who point this out often cite the ills of capitalism as the driving force behind this phenomena - however I’d like to point to something else as the driving force behind this phenomena… and that’s Moravec’s Paradox.

So, what’s Moravec’s Paradox?

Moravec’s paradox is the observation in artificial intelligence and robotics that, contrary to traditional assumptions, reasoning requires very little computation, but sensorimotor and perception skills require enormous computational resources. The principle was articulated by Hans Moravec, Rodney Brooks, Marvin Minsky and others in the 1980s.

One possible explanation of the paradox, offered by Moravec, is based on evolution. All human skills are implemented biologically, using machinery designed by the process of natural selection. In the course of their evolution, natural selection has tended to preserve design improvements and optimizations. The older a skill is, the more time natural selection has had to improve the design. Abstract thought developed only very recently, and consequently, we should not expect its implementation to be particularly efficient.

Moravec:

Encoded in the large, highly evolved sensory and motor portions of the human brain is a billion years of experience about the nature of the world and how to survive in it. The deliberate process we call abstract thought or reasoning is, I believe, the thinnest veneer of human thought, effective only because it is supported by this much older and much more powerful, though usually unconscious, sensorimotor knowledge.

We are all prodigious olympians in perceptual and motor areas, so good that we make the difficult look easy. Abstract thought, though, is a new trick, perhaps less than 100 thousand years old. We have not yet mastered it. It is not all that intrinsically difficult; it just seems so when we do it.

• Skills that developed recently in our evolutionary history such as speech, language, abstract thought etc may seem difficult but only because they are new & implemented in inefficient ways, this does however make them the easiest to reverse engineer or at least mimic (GPT-4, Midjourney, Deepmind etc). While things that seem trivial, like fine motor skills & manipulation of objects are traits that have been honed and refined for billions of years - not just hundreds of thousands - and as such are the most difficult to reverse engineer.

Moravec wrote in 1988, “it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility”.

Similarly, Minsky emphasized that the most difficult human skills to reverse engineer are those that are below the level of conscious awareness. “In general, we’re least aware of what our minds do best”, he wrote, and added “we’re more aware of simple processes that don’t work well than of complex ones that work flawlessly”. Similarly, Steven Pinker wrote in 1994 that “the main lesson of thirty-five years of Al research is that the hard problems are easy and the easy problems are hard.”

• Is an Interesting parallel to the recent development of transformers (feed-forward neural networks) in the space of art work (midjourney, stable diffusion, Dall-e) and language processing (GPT 3 - 4, LLaMA, Deepmind, LaMDA…etc)

In 2017, a leading machine learning researcher Andrew Ng presented a “highly imperfect rule of thumb”, that “almost anything a typical human can do with less than one second of mental thought, we can probably now or in the near future automate using Al.

In the 2020s, in accordance to Moore’s law, computers were hundreds of millions of times faster than in the 1970s & the additional computer power was finally sufficient to begin to handle perception and sensory skills, as Moravec had predicted in 1976. As seen in projects that include Boston Dynamics, things like Gemini, PaLM etc.

(Processing speed is relevant as thousands of years of human equivalent mental processing can occur in months on fast computers) - dubbed “speed intelligence” by Nick Bostrom. Neurons operate at a peak speed of 200 Hz - 14 times slower than a modern microprocessor- which operates at ~ 4 GHz

TLDR:

1. We should expect the difficulty of reverse-engineering any human skill to be roughly proportional to the amount of time that skill has been evolving in animals. (1.5 billion-ish years?)

2. The oldest human skills are largely unconscious and so appear to us to be effortless.

3. Therefore, we should expect skills that appear effortless to be difficult to reverse-engineer (physical labor) but skills that require effort may not necessarily be difficult to engineer at all (recent developments in AI art & language processing).