Forget the legend of the solitary scientist scribbling on a chalkboard at 3 a.m. until the universe coughs up its secrets. That story sells movie tickets, but in a new Neuron interview, neuroscientist Danilo Bzdok makes the case that human intuition - the very thing we've celebrated for centuries - has quietly become the bottleneck in brain science. The hero of the next era isn't a genius with a hunch. It's a machine that's read everything.

It's the scientific equivalent of a coach finally admitting the star player can't carry the whole offense anymore. Time to change the playbook.

The Brain Beat Has a Roster Problem

Here's the situation on the court. Neuroscience has spent a century drafting brilliant specialists - one expert on the hippocampus, another on dopamine, a third who has devoted her life to a single ion channel. Each is elite in their lane. The problem? They barely pass the ball. Knowledge piles up in isolated silos, and nobody has the time, or frankly the lifespan, to read all of it. Estimates put neuroscience output at a paper roughly every couple of minutes. No human is boxing that out.

Bzdok's argument is that the categories we've been using to organize the brain - the tidy little boxes labeled "memory," "emotion," "attention" - were built by smart people guessing. Useful guesses! But guesses, drawn from intuition rather than the firehose of actual data. And when your filing system is built on vibes, you eventually start missing the connections that matter.

Enter the Rookie Who Read the Entire Library

This is where large language models check into the game. The pitch isn't that an AI will out-think a brilliant neuroscientist on any single play. It's that an LLM can digest more research than any human possibly could, then spot patterns hiding in the gaps between specialties - the assist nobody on the floor could see.

Think of it as the ultimate utility player. It doesn't get tired in the fourth quarter. It read the paper on cortical folding AND the one on gut bacteria AND the obscure 1987 study everyone forgot, and it noticed they might be talking about the same thing. Bzdok and colleagues laid out the playbook for this in an earlier Neuron piece, arguing LLMs are a genuinely new asset class for science, not just a fancier search bar (Bzdok et al., 2024).

And this isn't hype talking. In one study, LLMs flat-out beat human experts at predicting how neuroscience experiments would turn out (Luo et al., 2024). Let that sit for a second. The machine called the game better than the people who play it for a living. That's not a stat you bury in the box score.

The Culture Has to Change, Not Just the Tech

Now for the part that's less highlight reel and more locker-room reality. Bzdok is clear that better algorithms alone won't win the championship. Neuroscience needs a cultural shift toward treating data as the main event - the kind of change where AI-heavy studies get the prime slots in top journals and the big grants, instead of being treated as a sideshow to the "real" lab work.

That's a tough sell in a field that still romanticizes the lone thinker. Asking a discipline to bench its founding myth is like telling a team that's always run the same offense that the whole system needs a rebuild. Nobody loves hearing it. But the data-driven approach is already racking up wins elsewhere - massive resources like the UK Biobank let researchers predict age, cognition, and mental health straight from brain scans across hundreds of thousands of people, a scale no intuition-led study could ever touch (Li et al., 2026).

Why This Matters Past the Final Buzzer

If Bzdok is right, the payoff is huge. A unified map of brain function - one that finally stitches the silos together - could reshape how we understand and treat everything from depression to dementia. Diseases don't respect our tidy categories, so maybe our science shouldn't either.

The lone genius isn't getting cut from the team. Human creativity still calls the plays and asks the questions that matter. But the next era of brain science looks less like a one-on-one showdown and more like a well-drilled squad where the smartest move is knowing when to pass to the machine. The buzzer on the old way is sounding. Game on.

Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.

References

1. Bzdok, D. (2026). Danilo Bzdok [Interview]. Neuron. DOI: 10.1016/j.neuron.2026.04.042. PMID: 42235489

2. Bzdok, D., et al. (2024). Data science opportunities of large language models for neuroscience and biomedicine. Neuron, 112(5), 698-717. DOI: 10.1016/j.neuron.2024.01.016. PMID: 38340718

3. Luo, X., et al. (2024). Large language models surpass human experts in predicting neuroscience results. Nature Human Behaviour. DOI: 10.1038/s41562-024-02046-9. PMCID: PMC11860209

4. Li, Y., Gao, H., Lin, L., Wu, Y., & Zhu, X. (2026). UK Biobank-centric advances in brain age prediction: a comprehensive review. Reviews in the Neurosciences. DOI: 10.1515/revneuro-2025-0055