Polanyi's paradox
In plain wordsThe observation that we know how to do many things — recognize a face, ride a bike — that we cannot explain in words. That made such skills hard to program until machines learned them from examples.
The observation that much human knowledge is tacit: people perform tasks such as recognizing a face, judging a situation or riding a bicycle without being able to state the rules they follow. Applied to automation, it explains why tasks people do effortlessly long resisted computerization: programmers cannot write down procedures that no one can articulate, so early automation concentrated on explicit, routine tasks. The paradox was a mainstay of criticism of symbolic AIsymbolic AI, n.: An approach to AI that writes knowledge down as human-readable symbols and rules, such as 'every bird has wings', and reasons by manipulating them step by step instead of learning patterns from data. and expert systemsexpert system, n.: A program that stores a human specialist's know-how as many if–then rules and uses them to give advice or make decisions in a narrow field, such as diagnosing infections or configuring computers., whose knowledge engineers struggled to elicit experts' know-how. Machine learningmachine learning, n.: A way of building software that learns from examples instead of following rules written by a programmer. Shown thousands of labeled photos, for instance, a program can learn to recognize cats. partly circumvents it by inferring tacit regularities from examples rather than from stated rules, which is why perception, translation and other once-intractable tasks became automatable in the 2010s; the knowledge so acquired is often just as tacit inside the model, a problem for interpretabilityinterpretability, n.: The effort to understand how an AI system arrives at its outputs — what information it uses and how it combines it — instead of treating it as a sealed black box..
Nobody could write the rules for spotting a tumor, which is Polanyi's paradox; the network learned them from labeled scans.
Origin
Named by the economist David Autor in 'Polanyi's Paradox and the Shape of Employment Growth' (2014), after the chemist and philosopher Michael Polanyi, whose The Tacit Dimension (1966) stated that 'we can know more than we can tell'.
Usage note
Related to but distinct from Moravec's paradoxMoravec's paradox, n.: The observation that what is hard for people, like chess or algebra, is often easy for computers, while what toddlers do effortlessly, such as seeing, walking and picking things up, is very hard for machines., which concerns the computational difficulty of sensorimotor skills rather than the inexpressibility of knowledge.
Related
- Moravec's paradoxMoravec's paradox, n.: The observation that what is hard for people, like chess or algebra, is often easy for computers, while what toddlers do effortlessly, such as seeing, walking and picking things up, is very hard for machines.
- expert systemexpert system, n.: A program that stores a human specialist's know-how as many if–then rules and uses them to give advice or make decisions in a narrow field, such as diagnosing infections or configuring computers.
- Dreyfus's critique of AIDreyfus's critique of AI, n.: The philosopher Hubert Dreyfus's argument, from the 1960s on, that rule-based AI could never match human intelligence, because much human skill rests on bodily know-how and intuition rather than explicit rules.
- automationautomation, n.: Using machines or software to do work that people would otherwise do, from factory robots to programs that sort email or write reports.
- machine learningmachine learning, n.: A way of building software that learns from examples instead of following rules written by a programmer. Shown thousands of labeled photos, for instance, a program can learn to recognize cats.
- interpretabilityinterpretability, n.: The effort to understand how an AI system arrives at its outputs — what information it uses and how it combines it — instead of treating it as a sealed black box.
- Moravec's paradoxMoravec's paradox, n.: The observation that what is hard for people, like chess or algebra, is often easy for computers, while what toddlers do effortlessly, such as seeing, walking and picking things up, is very hard for machines.
Sources
- Polanyi, M. (1966). The Tacit Dimension. Doubleday.
- Autor, D. H. (2014). Polanyi's Paradox and the Shape of Employment Growth. NBER Working Paper; Federal Reserve Bank of Kansas City Economic Policy Symposium, Jackson Hole.
Mentioned in
frame problemframe problem, n.: The difficulty of getting an AI to work out what does and does not change when something happens, without spelling out every one of the countless things that stay the same., Moravec's paradoxMoravec's paradox, n.: The observation that what is hard for people, like chess or algebra, is often easy for computers, while what toddlers do effortlessly, such as seeing, walking and picking things up, is very hard for machines., Dreyfus's critique of AIDreyfus's critique of AI, n.: The philosopher Hubert Dreyfus's argument, from the 1960s on, that rule-based AI could never match human intelligence, because much human skill rests on bodily know-how and intuition rather than explicit rules., factored cognitionfactored cognition, n.: The idea that a hard thinking task can be broken into many small, separately checkable pieces, each simple enough for a person or an AI to handle in a short time., Lovelace objectionLovelace objection, n.: The argument, traced to Ada Lovelace in 1843, that computers cannot be truly creative because they can only do what they have been programmed to do.