Moravec's Paradox

👇 Get new sketches each week

On the capabilities of machines and people

In his 1988 book Mind Children , Austrian robotics researcher Hans Moravec, wrote :

“…as the number of demonstrations has mounted, it has become clear that it is comparatively easy to make computers exhibit adult-level performance in solving problems such as intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception or mobility.”

This surprising pattern became known as Moravec’s Paradox, the idea that machines often do well at things people find hard, and do poorly at what is easy for us.

As classic examples:

We consider chess grandmasters to be among the most intelligent people on the planet, yet computers eventually triumphed over them.
A young child can (in principle) place raspberries gently on a plate, stack glasses in a dishwasher, read a friend's emotions, crawl up some stairs, or play with a dog, and yet our machines still find this hard.

It sounds rather reasonable.

That Google Maps can, in seconds, provide me with an optimised walking route between places hundreds of miles apart is a source of consistent amazement for me. Yet it’s challenging for a self-driving car to understand if a traffic official is gesturing for it to slow down, something I find effortless.

Yet some of the traditional examples of Moravec’s paradox now feel dated. As Hans said, “Unfortunately for humanlike robots, computers are at their worst trying to do the things most natural to humans, such as seeing, hearing, manipulating objects, learning languages, and commonsense reasoning.” Machines have now made great strides in most of these.

Princeton Computer Science Professor Arvind Narayanan, thinks Moravec’s Paradox is neither useful nor true .

Yet UC Berkeley Robotics Professor Ken Goldberg cautions that robotics still faces a “100,000 year data gap .” He explains that the extraordinary capabilities of models that produced the AI capabilities we all know are created from the internet-scale amount of image and text data we already had available. He calculated that it would take a person 100,000 years to read or view all the data that we had ready to feed into our AI models.

In contrast, we have comparatively little data ready to use for applying similar techniques to, say, robot dexterity. He advocates an approach where we make robots that are functional enough to be useful, and use these for data collection to improve, much as Waymo have been doing with self-driving cars for over 15 years.

What do you think?

Related Ideas to Moravec’s Paradox

Also see:

The Automation Paradox: the better the machines get, the more we struggle when they fail
Fauxtomation: claims of automation for work actually done by people
Cognitive offloading: delegating our thinking
Jevon’s Paradox: Fuel efficiency gains tend to increase, not decrease fuel use
Chaos monkey
Looking under the Lamppost
More paradoxes

Ken Goldberg’s view on robotics at UC Berkeley News, Aug 2025

✨ Latest

Get the book!

Big Ideas Little Pictures

Prefer to listen?

Try the podcast

🎉 Help keep Sketchplanations ad-free 🎉 Keep this ad-free 🎉 Help keep Sketchplanations ad-free by supporting me

Keep exploring

Pace Layers: Six layers of robust and adaptable civilisations

A world of 4 income levels

Laplace's Demon

Quote and advice from Jane Goodall's mother: If there's something you really want, you're going to have to work really hard, take advantage of opportunity and, above all, never give up.

And, above all, never give up

Chiasmus

The three-day effect: spending three days in nature improves creativity, clarity and wellbeing

The 3-day effect

What makes a resilient system? It needs to be strong, but flexible. Pace Layers, sometimes called pace layering, is a framework created by Stewart Brand of the Long Now Foundation that explores how different parts of civilisation change at different speeds. Fast-moving layers like fashion bring novelty and experimentation, while slower layers like culture and nature provide stability and memory. Together, the layers support, reinforce, and challenge each other—creating robust, adaptable societies. Pace layers is a model I have come back to again and again. What is the Pace Layers Framework? The pace layering model has six levels, with each deeper layer moving more slowly than the one above it. In Brand's words: Fast learns, slow remembers. Fast proposes, slow disposes. Fast is discontinuous, slow is continuous. Fast and small instructs slow and big by accrued innovation and occasional revolution. Slow and big controls small and fast by constraint and constancy. Fast gets all our attention, slow has all the power. In short: the fast layers innovate, the slow layers stabilise. Brand argued that all durable dynamic systems have this sort of structure—it’s what makes them adaptable and resilient. According to pace layering, the six layers of pace and size in the working structure of a robust and adaptable civilisation are: Fashion/Art Commerce Infrastructure Governance Culture Nature Pace Layering Example: Electric Scooters An example brings pace layers home to me. Here, I consider electric scooters, but you can take almost any innovation and think through the layers. Fashion/Art New scooter designs, colours, marketing campaigns, and influencer hype cycles. These come and go quickly—some stick, most fade. Commerce Scooter rental startups, pricing models, venture capital funding, and competition with bikes, cars, and other transport. Commerce runs slower than fashion, but still moves quickly through experimentation. Infrastructure Charging docks, scooter lanes, designated parking areas, and integration with city planning—all these take time. Cities and transport systems have to adapt, if electric scooters persist, for them to thrive rather than die out. Governance The laws and regulations surrounding scooters often lag behind the pace of commerce, including speed limits, taxes, penalties, incentives, safety requirements, age restrictions, or bans. Governance responds as the effects of scooters unfold. Culture Culture in the deep sense of how this changes mobility and ideas around travel. It may affect socialising, where people work and live, and what we choose to do as a society. Culture evolves slowly as habits and norms change in tandem with governance and infrastructure. Nature The environmental realities of scooters vs other transport means. Batteries, natural resources, long-term sustainability, and the effect on the climate. Applying the Pace Layers Model to AI Applying pace layers to the growth and development of Artificial Intelligence (AI) is also instructive: Fashion: AI tools and interfaces change daily. Commerce: Models and capabilities change weekly or monthly. Companies compete on quarterly and annual cycles. Infrastructure: Investment, data centres and energy generation are being approved and built slowly. Governance: The rules, ethics, court cases, and legal challenges surrounding the regulation of AI content and companies are developing on a case-by-case basis. Culture: How AI changes work, employment, where people live, and power structures are all in progress. Nature: Effects on the planet, energy and climate take longer to play out. AI can feel like it's changing every day, yet its deepest consequences will play out over decades and centuries. Pace Layers and Time Hierarchy In his book The Clock of the Long Now, Brand also discusses time hierarchy in nature. He describes a coniferous forest: The needle changes within a year. The tree crown over several years. The patch over many decades. The stand over a couple of centuries. The forest over a thousand years. The biome over ten thousand years. The faster elements are constrained by the slower ones: the needle by the crown, the crown by the patch, and so on up to the biome. Yet innovation still percolates through the system via evolutionary competition. Occasionally, shocks—fire, disease, human activity—disrupt the whole structure, sometimes all the way to the biome. You can see this idea in my sketch on Time Hierarchy. Considering the Pace Layers Model I find pace layers thinking interesting as a way to differentiate between the fast and slower changes I see around me. It helps me make better sense of the world and see the age-old tussle between innovation and stability. Related Ideas to Pace Layers Also see: Time Hierarchy The Overview Effect The Overton Window: the range of politically acceptable ideas Human pace The Gartner Hype Cycle Marchetti's Constant The Destiny Instinct The S-Curve The Law of Unintended Consequences I revised my original pace layers sketch because I kept discussing it with people and thought it needed a better visual. Read more on Pace Layers at the Long NowWhat makes a resilient system? It needs to be strong, but flexible. Pace Layers, sometimes called pace layering, is a framework created by Stewart Brand of the Long Now Foundation that explores how different parts of civilisation change at different speeds. Fast-moving layers like fashion bring novelty and experimentation, while slower layers like culture and nature provide stability and memory. Together, the layers support, reinforce, and challenge each other—creating robust, adaptable societies. Pace layers is a model I have come back to again and again. What is the Pace Layers Framework? The pace layering model has six levels, with each deeper layer moving more slowly than the one above it. In Brand's words: Fast learns, slow remembers. Fast proposes, slow disposes. Fast is discontinuous, slow is continuous. Fast and small instructs slow and big by accrued innovation and occasional revolution. Slow and big controls small and fast by constraint and constancy. Fast gets all our attention, slow has all the power. In short: the fast layers innovate, the slow layers stabilise. Brand argued that all durable dynamic systems have this sort of structure—it’s what makes them adaptable and resilient. According to pace layering, the six layers of pace and size in the working structure of a robust and adaptable civilisation are: Fashion/Art Commerce Infrastructure Governance Culture Nature Pace Layering Example: Electric Scooters An example brings pace layers home to me. Here, I consider electric scooters, but you can take almost any innovation and think through the layers. Fashion/Art New scooter designs, colours, marketing campaigns, and influencer hype cycles. These come and go quickly—some stick, most fade. Commerce Scooter rental startups, pricing models, venture capital funding, and competition with bikes, cars, and other transport. Commerce runs slower than fashion, but still moves quickly through experimentation. Infrastructure Charging docks, scooter lanes, designated parking areas, and integration with city planning—all these take time. Cities and transport systems have to adapt, if electric scooters persist, for them to thrive rather than die out. Governance The laws and regulations surrounding scooters often lag behind the pace of commerce, including speed limits, taxes, penalties, incentives, safety requirements, age restrictions, or bans. Governance responds as the effects of scooters unfold. Culture Culture in the deep sense of how this changes mobility and ideas around travel. It may affect socialising, where people work and live, and what we choose to do as a society. Culture evolves slowly as habits and norms change in tandem with governance and infrastructure. Nature The environmental realities of scooters vs other transport means. Batteries, natural resources, long-term sustainability, and the effect on the climate. Applying the Pace Layers Model to AI Applying pace layers to the growth and development of Artificial Intelligence (AI) is also instructive: Fashion: AI tools and interfaces change daily. Commerce: Models and capabilities change weekly or monthly. Companies compete on quarterly and annual cycles. Infrastructure: Investment, data centres and energy generation are being approved and built slowly. Governance: The rules, ethics, court cases, and legal challenges surrounding the regulation of AI content and companies are developing on a case-by-case basis. Culture: How AI changes work, employment, where people live, and power structures are all in progress. Nature: Effects on the planet, energy and climate take longer to play out. AI can feel like it's changing every day, yet its deepest consequences will play out over decades and centuries. Pace Layers and Time Hierarchy In his book The Clock of the Long Now, Brand also discusses time hierarchy in nature. He describes a coniferous forest: The needle changes within a year. The tree crown over several years. The patch over many decades. The stand over a couple of centuries. The forest over a thousand years. The biome over ten thousand years. The faster elements are constrained by the slower ones: the needle by the crown, the crown by the patch, and so on up to the biome. Yet innovation still percolates through the system via evolutionary competition. Occasionally, shocks—fire, disease, human activity—disrupt the whole structure, sometimes all the way to the biome. You can see this idea in my sketch on Time Hierarchy. Considering the Pace Layers Model I find pace layers thinking interesting as a way to differentiate between the fast and slower changes I see around me. It helps me make better sense of the world and see the age-old tussle between innovation and stability. Related Ideas to Pace Layers Also see: Time Hierarchy The Overview Effect The Overton Window: the range of politically acceptable ideas Human pace The Gartner Hype Cycle Marchetti's Constant The Destiny Instinct The S-Curve The Law of Unintended Consequences I revised my original pace layers sketch because I kept discussing it with people and thought it needed a better visual. Read more on Pace Layers at the Long Now

Newer

RRandom

Older

As an Amazon Associate I earn from qualifying purchases.