Moravec’s Paradox is an observation made by roboticist Hans Moravec in 1988 that high-level reasoning requires very little computation, while low-level sensorimotor skills require enormous computational resources. It highlights how tasks humans find difficult, such as complex math, are easy for computers. On the other hand, things we find easy, like walking through a room or sensing face expressions, remain incredibly challenging for artificial intelligence systems to copy.
This concept is vital because it explains the current gap between digital intelligence and physical robotics. It reveals that basic skills like perception and movement are ancient, shared with animals and refined over millions of years of evolution, making them deeply ingrained and complex. Conversely, reasoning and math are relatively recent human developments, suggesting they may be less computationally complex and therefore easier to mimic by AI.
Some critics argue that Moravec’s Paradox is becoming less relevant due to modern scaling laws and deep learning. They suggest that with enough data and computer power, the hard problems of perception are finally being solved through massive neural networks. Others believe the issue was always about a lack of diverse training data rather than a fundamental limitation of machine logic or the inherent complexity of human evolution.
Regarding work automation, Moravec’s Paradox suggests that cognitive office roles might be automated faster than manual labor. AI agents can manage schedules or draft contracts but struggle with physical tasks in messy settings. The consequence of Moravec’s Paradox is that knowledge work will be automated much earlier than blue-collar roles or skilled trades that require complex physical interaction with the world.
