Neuroscience as AI Infrastructure Glossary
Neuroscience as AI infrastructure is the proposition that shared tools for measuring, mapping, and modelling brains can become foundations for AI research, as well as for understanding biological intelligence.
Infrastructure is the work that lets other work happen: instruments, datasets, standards, and reproducible methods. A connectome maps neural connections; recordings capture aspects of activity; predictive models test whether an account of that activity holds up on new observations. These resources give researchers something more demanding than an attractive analogy between a brain and a computer.
The relationship runs both ways. AI can help reconstruct neural wiring and analyze recordings. Neuroscience may suggest learning rules, architectures, or experimental questions for AI. Neither direction implies that a wiring diagram is a working mind, or that copying biology is the only route to capable machines.
This is a Dictionary framing, prompted by our discussion of Adam Marblestone’s research-infrastructure agenda, rather than the name of an established technical discipline. Its central question is practical: what shared measurement or tool would let many researchers test ideas they currently cannot test? Better measurement creates opportunities; it does not guarantee a breakthrough.
See also
Background
-
Adam H. Marblestone, Greg Wayne, and Konrad P. Kording, Toward an Integration of Deep Learning and Neuroscience, Frontiers in Computational Neuroscience (2016).
-
Adam Marblestone, AI is missing something fundamental about the brain, interview with Dwarkesh Patel. This conversation prompted the Dictionary framing; the term is our own.