A few days ago, reading Calum Chace’s exciting book “The Economic Singularity: Artificial Intelligence and the Death of capitalism,” I came across two concepts that I found interesting and on which I would like to reflect today. These are Moravec’s Paradox and Tesler’s Theorem.
The Moravec paradox
It is one of the most intriguing concepts in artificial intelligence (AI). Proposed by Hans Moravec, a pioneering robotics scientist, this paradox holds that it is easier for machines to perform tasks that we find intellectually challenging, such as chess or mathematical calculation, than those that seem simple and automatic to us, such as walking, picking up a bag from the floor or tying shoelaces.
Why does this happen?
Evolution has honed our motor skills over millions of years. These skills, which we consider fundamental, result from complex neural processes. On the other hand, tasks such as mathematical calculation the game of chess, or other types of games that are relatively recent in our evolutionary history, are based on clear and defined rules that can be easily programmed in a machine.
The challenge of teaching AI “simple” tasks
Although it may seem counterintuitive, teaching a machine to tie its laces is a monumental challenge. It involves motor coordination, tactile perception, and adaptability to different types of shoes and laces. It’s a perfect example of how something we take for granted can be incredibly complex from a programming standpoint.
Tesler's theorem and its relation to Moravec's paradox
“AI is whatever hasn’t been done yet.”
-Larry Tesler
While Moravec’s paradox focuses on the relative difficulty of programming certain human abilities into a machine, Tesler’s theorem offers another perspective on the complexity of programming itself. Larry Tesler, a well-known computer scientist, proposed the theorem that “every application has an inherent complexity that cannot be reduced”.
Basically, Tesler’s theorem suggests that there will always be a level of complexity that cannot be eliminated, regardless of how much we try to simplify or automate a process. In the context of AI, this means that there will always be challenges that require human intervention or decisions based on human judgment.
Two exciting artificial intelligence terms that I hope you have found interesting.
Good week!
