It is difficult or impossible to offer computer the talents of a one-year-old and easy to form computers exhibit adult level performance.”
Reverse-engineering a person's skill was expected by AI practitioners to be proportional to the quantity of your time that skill had been evolving during mankind's evolution. In the early stages of AI, skills that appeared effortless were expected to be difficult to reverse-engineer (typically cognitive functions like reading comprehension, visual perception , speech recognition). Teaching a machine to beat a human playing chess has been considered comparably easy by AI practitioners as it is a skill that only requires effort (mainly compute power).
The Moravec’s paradox is a factor that held back developments in AI. The availability of pretrained cognitive functions by hyperscale cloud providers is breaking it and fueling the current wave of AI.
The Moravec Paradox was assuming the following:
Oldest human skills are largely unconscious, so appear to us to be effortless.
Therefore, Skills that require effort may not necessarily be difficult to engineer at all but we should expect skills that appear effortless to be difficult to reverse-engineer,