symbolic AI
An approach to AI based on humans programming explicit, human-interpretable rules for behavior, rather than learning from data; favored by Minsky and Papert.
As defined in 68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
Other definitions
An early approach to artificial intelligence that tried to embody thinking in logical symbols and explicit rules for transforming them, treating intelligence as manipulation of formal logic.
Thinking too logically can actually hold you back - Steve StouteFounder and CEO, UnitedMasters and Translation
Where it comes up · 11
- Don’t Believe AI Hype, This is Where it’s Actually Headed | Oxford’s Michael Wooldridge | AI History12 claims
- Geoff Hinton - Will Digital Intelligence Replace Biological Intelligence? | Vector's Remarkable 20247 claims
- Human vs. Machine Consciousness | Imperial’s Murray Shanahan4 claims
- Thinking too logically can actually hold you back - Steve StouteFounder and CEO, UnitedMasters and Translation4 claims
- Will AI outsmart human intelligence? - with 'Godfather of AI' Geoffrey Hinton3 claims
- 280 | François Chollet on Deep Learning and the Meaning of Intelligence3 claims
- 68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense3 claims
- 319 - The Digital Multiverse2 claims
and 3 more
Most important claims · 10 of 41Search all claims for “symbolic AI” →
- Around 1970 Marvin Minsky and Seymour Papert mathematically proved that perceptrons were fundamentally limited in what they could learn no matter how much training data was provided, partly motivated by viewing neural networks as a competitor to their preferred symbolic AI approach.68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
- Symbolic AI researchers rejected Hinton's predictive word-vector approach as a 'silly way' to solve language problems, arguing that searching discrete symbol-manipulation rules was superior, but as the scale of problems increased, the feature-interaction approach proved vastly more scalable and effective than symbolic rule-based methodsGeoff Hinton - Will Digital Intelligence Replace Biological Intelligence? | Vector's Remarkable 2024
- Minsky and Papert's 1969 book Perceptrons proved that single-layer neural networks cannot solve the XOR problem, and by insinuating without proof that multi-layer networks also could not, they effectively killed neural network research and triggered the first AI winter, possibly motivated by their preference for symbolic AI funding.336 | Anil Ananthaswamy on the Mathematics of Neural Nets and AI
- When a programmer writes an exact static program describing a task so precisely that no uncertainty or novelty remains, the resulting program—though people call it an AI—contains no intelligence; the only intelligence in the picture is the mind of the programmer who understood and modeled the problem.280 | François Chollet on Deep Learning and the Meaning of Intelligence
- Herbert Simon and Allen Newell, two founding fathers of AI, built the 'general problem solver,' which because computers were so expensive was first codified on paper and executed by hand—one of them reportedly having his family help simulate how a computer would solve complex problems.Thinking too logically can actually hold you back - Steve StouteFounder and CEO, UnitedMasters and Translation
- Symbolic AI worked well on simple toy problems but failed as problems grew complex because the search space of possible solutions became enormous; representing real-world problems as logic problems caused the systems to break down once they moved beyond toy cases.Thinking too logically can actually hold you back - Steve StouteFounder and CEO, UnitedMasters and Translation
- Symbolic AI researchers criticized Hinton's 1985 language model approach as a silly way to solve language problems, noting that inductive logic programming could produce similar results by directly searching discrete rule spaces; but as scale increased, the neural approach proved far superior to rule-based manipulation.Geoff Hinton - Will Digital Intelligence Replace Biological Intelligence? | Vector's Remarkable 2024
- Early AI theorists adopted Symbolic AI—the idea that thinking could be embodied in logical symbols and explicit rules for transforming them—which forms a clear through line from Socrates's project of reducing knowledge to explicit definitions through the Enlightenment to AI's founding at Dartmouth in the 1950s.Thinking too logically can actually hold you back - Steve StouteFounder and CEO, UnitedMasters and Translation
- Knowledge-representation/symbolic AI approaches of the 1960s-70s failed because we couldn't represent the world in logical terms, and non-symbolic machine learning has given far better results, but we are now running up against the very limitations symbolic AI proponents warned about—that these systems are hyper black boxes acting merely as mirrors synthesizing the human-created data fed into them.319 - The Digital Multiverse
- The fundamental dichotomy in AI is not symbolic versus connectionist, but between programmers hard-coding a symbolic model of a task versus systems that learn from data how to perform a task; symbolic approaches are more tractable with limited compute, while learning approaches require more computational resources.280 | François Chollet on Deep Learning and the Meaning of Intelligence