https://www.cell.com/neuron/pdf/S0896-6273(17)30509-3.pdf

Abstract

The fields of neuroscience and artificial intelligence (AI) have a long and intertwined history. In more recent times, however, communication and collaboration between the two fields has become less commonplace. In this article, we argue that better understanding biological brains could play a vital role in building intelligent machines. We survey historical interactions between the AI and neuroscience fields and emphasize current advances in AI that have been inspired by the study of neural computation in humans and other animals. We conclude by highlighting shared themes that may be key for advancing future research in both fields.

Notes

• The interaction between neuroscience and AI has become less common, but the review emphasizes the ongoing importance of neuroscience in guiding AI research.
• Building human-level general AI is challenging due to the vast search space of possible solutions.
• Studying the human brain provides proof that human-level intelligence is possible and offers valuable insights.
• Animal cognition and its neural implementation can provide insights into higher-level general intelligence.
• Neuroscience offers inspiration for new algorithms and architectures that complement traditional AI approaches.
• Validating AI techniques through neuroscience can enhance their plausibility as components of general intelligence.
• Biological plausibility serves as a guide but is not strictly required from an engineering perspective.
• The focus is on understanding the algorithms, architectures, functions, and representations used by the brain.
• The approach differs from initiatives that aim to mimic or reverse-engineer neural circuits.
• Insights into general mechanisms of brain function can be gained while accounting for the unique challenges of building intelligent machines.
• The terms "neuroscience" and "AI" are used broadly, encompassing the study of the brain, behaviors, mechanisms, machine learning, statistics, and building intelligent machines.
• Deep learning and reinforcement learning, key areas of AI research, have roots in ideas from neuroscience.
• Current AI research has drawn inspiration from neuroscience, often without explicit acknowledgment.
• Deep learning has been combined with other machine learning approaches, such as reinforcement learning and Monte Carlo tree search.
• The potential for neuroscience to support future AI research is explored, along with emerging neuroscience-inspired AI techniques.
• The article also considers ways in which AI can benefit neuroscience and the potential for synergistic interactions between the two fields.