Split: The New Frontier of AI and Neuroscience

In the sun-drenched coastal city of Split, Croatia, an intellectual alchemy is quietly unfolding, far from the frenetic pace and billion-dollar valuations of Silicon Valley.

While venture capitalists continue to pour unprecedented sums—a staggering $110 billion in 2024 alone—into the latest AI applications, from automated hedge funds to dating apps, a different kind of conversation is taking root.

Here, amidst the ancient Roman squares and the shimmering Adriatic, a small but profoundly influential gathering is tackling questions more fundamental than the next incremental upgrade: How do we think? How do machines learn? And what happens when these two profound mysteries begin to converge?

This was the scene at the sixth International Conference on Mathematics of Neuroscience and AI, a four-day convocation in May 2025 that one attendee aptly described as “neurIPS crossed with Burning Man.”

It was anything but a typical academic conference.

Instead, it was a crucible for researchers, entrepreneurs, and visionaries, united by a shared conviction that the current AI boom, built largely on scaled-up neural networks and massive datasets, may be approaching a fundamental limit.

The paradox is stark: immense capital flows into building on top of existing AI models, yet comparatively little is invested in the basic research that could unlock truly next-generation intelligent systems. Funding for AI research is crucial.

Croatia, it seems, has become ground zero for wrestling with this very dilemma.

The conference’s origins trace back to a simple, yet audacious, idea six years ago, born from a frustration shared by Dr. Ruairidh McLennan Battleday.

A postdoctoral research fellow at Harvard’s Center for Brain Science and MIT’s Center for Brains, Minds, and Machines, Battleday observed a glaring void in neuroscience.

“There aren’t any truly general theories about how human brains learn and function,” he explained, noting that other sciences, like biochemistry, advanced by embracing more sophisticated mathematics.

What began as a gathering of Battleday, his colleagues Professor Dan Nicolau Jr and Dr. James Whittington, and a dozen friends, has since blossomed into a premier international forum, boasting what organizers confidently call “the best line-up of the decade for a neuroscience and AI conference.”

This year’s roster was indeed a who’s who of intelligence research, featuring luminaries like Professor Juergen Schmidhuber, widely acknowledged as the “father of modern AI” for his pioneering work in the early 1990s, and Professor Jay McClelland, one of the original architects of artificial neural networks.

The conference’s unwavering focus on mathematical rigor underscores a growing realization: neuroscience, despite its century of existence, still lacks the unifying theoretical frameworks that define fields like physics.

“We’re still trying to figure out the basic rules,” Battleday observed, emphasizing the profound implications of such discoveries—from understanding ourselves better to treating neurological diseases and, crucially, building far more capable and intelligent machines.

The interdisciplinary spirit permeated every aspect of the event.

Seven globally renowned keynote speakers delivered presentations spanning the theoretical frontiers of neuroscience to the practical applications of AI.

Another thirty talks and two hundred posters delved into everything from biocomputation to cognitive science, fostering an intellectual environment described by attendees as exceptionally rich and stimulating.

Schmidhuber, speaking to a packed auditorium, highlighted a critical gap: despite his own foundational contributions to modern AI algorithms, he argued that the physical means for these systems to effectively learn and develop in the real world are still largely absent.

Perhaps the most provocative moment came from Professor Jay McClelland of Stanford and DeepMind.

His keynote, “Are People Still Smarter than Machines?”, offered a nuanced, qualified “yes.” While acknowledging AI’s astounding progress, McClelland posited that the human brain employs “a more efficient and creative algorithm that follows a different—and better—logic.”

This assertion challenges the prevailing narrative of AI’s inexorable march towards or beyond human-level intelligence in all domains.

McClelland’s argument suggests that the true frontier of AI may not be about scaling up existing models or amassing more data, but rather about fundamentally rethinking how information is processed and how creativity emerges, potentially mirroring human cognition more closely.

The implications are enormous, hinting at a future where AI development shifts from brute computational force to a more elegant, biologically inspired approach.

A striking feature of the Split conference was its seamless integration of academic rigor with real-world urgency.

The presence of industry leaders, including Erwann Le Lannou from XTX Markets, the conference’s main sponsor, alongside university researchers, signaled a significant evolution in how neuroscience and AI research is conducted and funded.

The atmosphere was one of “exceptional talent yet a relaxed vibe,” a conducive environment for “people to speak freely with each other and the great minds we attract.”

The organizers themselves stressed that “offline discussions are a key part of the conference—that’s where the really radical insights are proposed and debated.” Indeed, in the ancient squares of Split’s old town, over shared meals and leisurely walks along the Dalmatian coast, participants engaged in the kind of uninhibited intellectual exchange that has historically fueled scientific breakthroughs.

The timing of this unique gathering could not be more critical. As Battleday soberly noted, “Right now, with AI, it feels like we are on the verge of something great or disastrous.” The relentless march of AI, from sophisticated language models to nascent autonomous systems, presents humanity with both unprecedented opportunities and existential concerns.

The interdisciplinary ethos championed in Split offers a potential compass through this uncertain terrain.

By uniting neuroscientists, AI researchers, mathematicians, and industry practitioners, the event cultivates the cross-pollination necessary for truly transformative insights. Understanding the impact of AI on society may, in fact, be the key to developing AI systems that are not only powerful but also inherently safe and beneficial.

The convivial spirit, encapsulated by Dr. Ida Momennejad of Microsoft Research exclaiming, “I think I finally get neuromonster!”—a playful nod to the conference’s affectionate nickname—underscored the unique culture that has blossomed around this gathering.

As the conference drew to a close, with participants toasting their collective efforts and setting their sights on next year’s gathering in Rhodes, Greece, the profound questions posed in Split lingered.

The fact that the entire event was volunteer-run, organized by active researchers with no formal event-planning experience, speaks volumes about the genuine passion and shared mission driving this community.

The conference’s remarkable growth from a small circle of friends to a major international forum reflects a crucial realization: the traditional boundaries between neuroscience and AI are not merely blurring; they are becoming increasingly irrelevant.

As artificial systems grow ever more sophisticated and our understanding of biological intelligence deepens, the most pressing questions are not whether humans or machines are “smarter,” but rather how we can best harness both forms of intelligence to navigate the complex challenges ahead.

The dialogues initiated in Split’s Roman squares, continued on its coastal paths, and debated in its lecture halls, are poised to influence research trajectories, funding priorities, and technological development for years to come.

In an era where the pace of AI advancement shows no signs of abatement, forums like this one may prove indispensable in ensuring that progress ultimately serves humanity’s highest interests.

For now, the researchers who converged on Croatia have dispersed, armed with fresh perspectives and connections forged through four days of intense intellectual exchange.

The true measure of the conference’s impact will not be found in papers published or presentations delivered, but in the breakthroughs that emerge from the unique intersection of human curiosity and artificial capability that defines our current moment in history.