mlsanityData to 7 Oct 2026

Eight challenges in developing theory of intelligence

arXiv 2306.11232 ↗First posted 20 Jun 2023q-bio.NC cond-mat.stat-mech cs.AIcited by 6

Not on the map. Why?

Abstract (arXiv)

A good theory of mathematical beauty is more practical than any current observation, as new predictions of physical reality can be verified self-consistently. This belief applies to the current status of understanding deep neural networks including large language models and even the biological intelligence. Toy models provide a metaphor of physical reality, allowing mathematically formulating that reality (i.e., the so-called theory), which can be updated as more conjectures are justified or refuted. One does not need to pack all details into a model, but rather, more abstract models are constructed, as complex systems like brains or deep networks have many sloppy dimensions but much less stiff dimensions that strongly impact macroscopic observables. This kind of bottom-up mechanistic modeling is still promising in the modern era of understanding the natural or artificial intelligence. Here, we shed light on eight challenges in developing theory of intelligence following this theoretical paradigm. Theses challenges are representation learning, generalization, adversarial robustness, continual learning, causal learning, internal model of the brain, next-token prediction, and finally the mechanics of subjective experience.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 67 of its 146 references that are in the map. First posted on arXiv 20 Jun 2023.

Builds on

67 of 146 references.

References · 67

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2023GPT-4 Technical Reportcited by 18,132
  4. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  5. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  6. 2020Scaling Laws for Neural Language Modelscited by 6,411
  7. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  8. 2013Intriguing properties of neural networkscited by 5,114
  9. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  10. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  11. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
  12. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
38 more
  1. 2018World Modelscited by 1,444
  2. 2018Continual Lifelong Learning with Neural Networks: A Reviewcited by 1,313
  3. 2018Reconciling modern machine learning practice and the bias-variance trade-offcited by 919
  4. 2014Deep Learning in Neural Networks: An Overviewcited by 765
  5. 2017Opening the Black Box of Deep Neural Networks via Informationcited by 494
  6. 2020Hopfield Networks is All You Needcited by 461
  7. 2019Causality for Machine Learningcited by 242
  8. 2021Deep learning: a statistical viewpointcited by 239
  9. 2021Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolationcited by 130
  10. 2020Large Associative Memory Problem in Neurobiology and Machine Learningcited by 110
  11. 2018A high-bias, low-variance introduction to Machine Learning for physicistscited by 108
  12. 2017The Consciousness Priorcited by 105
  13. 2020The Neural Tangent Kernel in High Dimensions: Triple Descent and a Multi-Scale Theory of Generalizationcited by 101
  14. 2019Uncertainty-guided Continual Learning with Bayesian Neural Networkscited by 100
  15. 2021Continual Learning in the Teacher-Student Setup: Impact of Task Similaritycited by 86
  16. 2016Towards an integration of deep learning and neurosciencecited by 77
  17. 2018A jamming transition from under- to over-parametrization affects loss landscape and generalizationcited by 65
  18. 2022On the Principles of Parsimony and Self-Consistency for the Emergence of Intelligencecited by 50
  19. 2016Unreasonable Effectiveness of Learning Neural Networks: From Accessible States and Robust Ensembles to Basic Algorithmic Schemescited by 40
  20. 2022Large Language Models and the Reverse Turing Testcited by 38
  21. 2021Synaptic metaplasticity in binarized neural networkscited by 28
  22. 2023Brain-inspired learning in artificial neural networks: a reviewcited by 26
  23. 2021Interpreting neural computations by examining intrinsic and embedding dimensionality of neural activitycited by 24
  24. 2021A Theory of Consciousness from a Theoretical Computer Science Perspective: Insights from the Conscious Turing Machinecited by 22
  25. 2021Phase Transitions in Transfer Learning for High-Dimensional Perceptronscited by 21
  26. 2021Out-of-Distribution Generalization in Kernel Regressioncited by 21
  27. 2021Learning through atypical "phase transitions" in overparameterized neural networkscited by 21
  28. 2022Integrated information theory (IIT) 4.0: Formulating the properties of phenomenal existence in physical termscited by 21
  29. 2021Unified field theoretical approach to deep and recurrent neuronal networkscited by 20
  30. 2023Mapping of attention mechanisms to a generalized Potts modelcited by 19
  31. 2014Origin of the computational hardness for learning with binary synapsescited by 16
  32. 2022Supervised Hebbian Learningcited by 16
  33. 2020Mathematical Models of Overparameterized Neural Networkscited by 15
  34. 2019Minimal model of permutation symmetry in unsupervised learningcited by 13
  35. 2017A Separation Principle for Control in the Age of Deep Learningcited by 9
  36. 2022Statistical mechanics of continual learning: variational principle and mean-field potentialcited by 9
  37. 2020Will we ever have Conscious Machines?cited by 7
  38. 2019Statistical physics of unsupervised learning with prior knowledge in neural networkscited by 6

50 most cited of 67 All 67 (JSON)

Built on by

Cited by year

Citing papers per year: 2023: 1; 2024: 3; 2025: 2.

6 papers.

  1. 2025Neural Langevin Machine: a local asymmetric learning rule can be creativecited by 3
  2. 2023Meta predictive learning model of languages in neural circuitscited by 2
  3. 2024Fermi-Bose Machine achieves both generalization and adversarial robustnesscited by 2
  4. 2024Spin glass model of in-context learningcited by 2
  5. 2024AI-Empowered Human Research Integrating Brain Science and Social Sciences Insightscited by 0
  6. 2025Geometric origin of adversarial vulnerability in deep learningcited by 0