AI may give neuroscientists a new way to study how intelligence works, but David Sussillo thinks neuroscience is unlikely to reshape AI in the next few years. The computational neuroscientist argues that the fields can learn from one another, though the timescale matters.
Watch Desk analysis
What happened
In an interview with The Transmitter, Sussillo describes neuroscience as an early source of ideas for artificial neural networks, while arguing that modern AI has since moved ahead of neuroscience’s ability to explain it. He says neuroscience has not yet been especially successful at understanding AI systems, in part because the field’s theories are still developing and commercial AI changes quickly.
He sees a newer opportunity in studying trained models during inference, including their language-based reasoning, with behavioural approaches borrowed from cognitive science. Sussillo also points to practical routes between the fields: internships can help neuroscience graduates move into AI research, while stronger Python training would address a gap he sees in university programmes that rely on MATLAB.
Why it matters
This is a useful correction to the familiar idea that brain science is about to hand AI its next big breakthrough. Sussillo is more optimistic about AI helping neuroscience in the near term, because researchers can inspect and experiment on the whole artificial system in ways that are not possible with a brain.
His longer-term possibility is a large, high-quality neural-data model that could act as a kind of virtual brain for experiments. That is a substantial research ambition, not a tool he says exists today. For the nearer term, his message is more practical: AI research can benefit from neuroscientists’ expertise, even if neuroscience does not yet have a ready-made theory of how large AI systems work.
Our read
Sussillo’s most interesting point is that the exchange may currently run in the opposite direction from the popular story. AI gives brain researchers a new object to study, while neuroscience graduates bring useful skills into AI labs. The brain-inspired origin story is real; treating it as a blueprint for the next model would be a rather larger claim.
What to watch
- Whether researchers develop behavioural methods for studying AI models during inference.
- Whether AI firms continue hiring neuroscientists and cognitive neuroscientists.
- Whether a large foundation model of neural data becomes a credible research programme.
Discussion spark: Should AI researchers invest more in neuroscience now, or focus on understanding today’s models with the tools AI already provides?
Sources and evidence
- Source update (27 October 2026, 04:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.