The next paradigm: AI, Neural Networks, the Symbolic Brain

2022-07-12 · Source: computational biology blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Expert, quick

Summary

The author argues that current neural network (NN) paradigms, despite recent successes driven by increased computing power and data, are "moribund" and have hit a fundamental roadblock. They contend that the shift from "Good Old-Fashioned AI" (GOFAI), based on rules, symbols, and logic, to purely statistical NN models, has stifled true advancement since the late 1980s. The article proposes a new paradigm, termed the "symbolic brain," which aims to integrate symbol manipulation, logic, and mathematics, drawing inspiration from the human brain's capabilities, particularly language acquisition. This "vertical-horizontal model" envisions neurons as specialized microprocessors capable of building a symbolic brain, with a trilogy of papers and a new company planned to develop these concepts.

Key takeaway

For AI researchers and scientists evaluating the future direction of artificial intelligence, this perspective suggests that incremental improvements to existing neural network architectures may be insufficient. Consider exploring alternative computational models that re-integrate symbolic reasoning and logic, rather than solely relying on statistical approaches, to overcome current AI limitations and achieve more human-like cognition.

Key insights

Current neural networks are fundamentally limited; a new symbolic brain paradigm is needed.

Principles

• Brain performs symbol manipulation and logic.
• Neurons can act as specialized microprocessors.

Method

The proposed "vertical-horizontal model" describes neuron interaction where each neuron functions as a specialized microprocessor, enabling the construction of a symbolic brain.

Topics

• Neural Network Limitations
• Symbolic AI
• Cognitive Architectures
• Brain-Inspired Computing
• AI Paradigms

Best for: AI Researcher, AI Scientist, Research Scientist

Related on AIssential

• Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives — A neuro-symbolic architecture with a probabilistic loss efficiently learns to solve NP-hard problems from natural inputs.
• The Thinking Machine That Google and Anthropic will Ship Together — True AI cognition necessitates a new mathematical blueprint, moving beyond current LLM scaling and flat embeddings.
• AI may not need massive training data after all — Brain-inspired AI architectures can achieve human-like activity without massive training data, challenging data-centric development.
• Connecting Language and (Artificial) Intelligence: Princeton’s Tom Griffiths — Human and artificial intelligence can be understood through three complementary mathematical frameworks: rules/symbols, neural networks, and Bayesian probability.
• Your Brain Is the World’s Most Powerful Computer — And It Just Inspired a Revolution — Deep learning's power comes from mimicking the brain's layered, connected neuron structure to learn from data.
• Creating Intelligence: A Computational Foundation for AGI — A computational theory of mind grounds AGI in discrete set theory and topological plasticity for efficient, biologically-inspired cognition.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by computational biology blog.