Picture a future where artificial intelligence can predict scientific breakthroughs decades before human researchers discover them, or where machine learning reveals hidden patterns in brain tissue that have eluded scientists for generations. This isn’t science fiction – it’s happening right now.

The panel, “AI in Scientific Discovery,” brought together two pioneering researchers whose work exemplifies the transformative potential of AI-driven science. Moderated by Thiru Ramaraj (Assistant Professor at the School of Computing) who works at the intersection of computer science and biology, the discussion offered compelling glimpses into how AI is revolutionizing research across disciplines.

Mapping the Brain’s Hidden Architecture

Kaiwen Kam (Associate Professor of Cell Biology at Rosalind Franklin University of Medicine and Science) shared his remarkable journey into the brainstem; a region that controls some of our most vital functions like breathing, swallowing, and vocalization. Despite centuries of medical research, scientists only discovered key regions controlling breathing about thirty years ago, and many areas remain unmapped.

The challenge Kam faces is daunting: the brainstem’s reticular formation contains tens of thousands of neurons that all look frustratingly similar under traditional examination. There are no clear anatomical boundaries to guide researchers, making it like trying to find specific neighborhoods in a city where all the buildings look identical and there are no street signs.

Working with DePaul’s Ramaraj and Sourati, Kam’s team has turned to machine learning to analyze the expression patterns of tens of thousands of genes from the Allen Brain Institute’s massive mouse brain atlas. This publicly available dataset, funded by Microsoft co-founder Paul Allen, represents the kind of industrial-scale science that no single laboratory could accomplish alone. By applying sophisticated clustering algorithms to this genetic data, the team is beginning to parse this seemingly uniform brain region into distinct functional modules – potentially revealing the neural circuits that keep us alive with every breath we take.

Predicting Tomorrow’s Discoveries Today

While Kam uses AI to map uncharted biological territory, Jamshid Sourati (Assistant Professor in DePaul’s School of Computing) is pushing the boundaries of what AI can predict about scientific progress itself.

Sourati’s team has developed models that analyze millions of scientific publications, examining not just their content but the hidden patterns in how scientists collaborate and cite each other’s work. The insight is elegantly simple yet profound: if two scientific concepts have no researchers working on both simultaneously (no bridge of collaboration connecting them) then a relationship between them likely hasn’t been discovered yet.

Using this approach, Sourati’s algorithms have successfully predicted scientific discoveries up to twenty years before they actually occurred. In material science, the team identified properties of thousands of materials that wouldn’t be discovered by human scientists for decades. In medicine, they’ve uncovered potential new uses for already-approved drugs – a process called drug repurposing that could dramatically accelerate treatment development.

Perhaps most intriguingly, Sourati explained how these algorithms can be tuned not just to follow the collective path of scientific inquiry, but to deliberately avoid it – generating what he calls “alien hypotheses” that human scientists might never imagine on their own. This raises fascinating questions about the nature of discovery itself: can AI help us escape the cognitive constraints that limit human imagination?

The Human-AI Partnership in Science

Throughout the discussion, both speakers emphasized a crucial point: AI isn’t replacing scientists but augmenting their capabilities in unprecedented ways. As Kam noted, the hypotheses generated by machine learning still require careful experimental validation.

This partnership between human expertise and machine intelligence emerged as a central theme. Sourati explained that while his algorithms can operate independently once given the right inputs, human experts remain essential for identifying which questions to ask, which properties matter most, and whether the AI’s predictions make scientific sense. The relationship is symbiotic: AI can process vast amounts of data and identify patterns invisible to human perception, while humans provide the context, creativity, and critical evaluation that ensures these discoveries are meaningful.

Kam’s experience working with computer scientists has been transformative in unexpected ways. Beyond the obvious benefits of computational efficiency, he found that the collaboration introduced entirely new ways of thinking about data – an algorithmic perspective that complemented his biological intuition. This cross-pollination of ideas represents the best of interdisciplinary research, where the sum truly exceeds its parts.

Looking Ahead: Science in the Age of AI

The panel raised provocative questions about the future of scientific research. One audience member humorously suggested that AI might soon automate the entire grant proposal process – if algorithms can predict which discoveries will succeed, why not let them decide which projects to fund? While tongue-in-cheek, the question touches on serious considerations about how AI will reshape the scientific enterprise.

Another crucial discussion point centered on trust. How can domain experts learn to rely on AI-generated hypotheses, especially when these predictions venture into unexplored territory? The panelists suggested that building this trust requires a feedback loop: as AI predictions are experimentally validated, confidence grows. Fields with rapid experimentation capabilities, such as those using robotic laboratories, may see this trust develop more quickly.

For aspiring scientists, the message was clear: the future belongs to those who can bridge domains. Sourati advised that regardless of their primary field, young researchers should develop at least basic data science literacy. The ability to work with large datasets and understand statistical principles will become increasingly essential as AI-generated insights permeate every scientific discipline.

The Promise and Responsibility

What emerged from this panel was a vision of AI not as a replacement for human scientific inquiry, but as a powerful partner that can accelerate discovery, reveal hidden patterns, and even suggest possibilities beyond human imagination. The work of researchers like Kam and Sourati demonstrates that we’re already living in this future – where algorithms can peer decades ahead in scientific progress or illuminate the mysterious workings of our own brains.

Yet with this power comes responsibility. As these technologies advance, the scientific community must grapple with fundamental questions about the nature of discovery, the role of human creativity, and how to ensure that AI-augmented science serves humanity’s best interests. The collaboration between DePaul’s School of Computing and researchers across disciplines positions the university at the forefront of these crucial conversations.

The AI revolution in scientific discovery isn’t coming – it’s here. And as this panel made abundantly clear, the most exciting breakthroughs will come not from AI or human intelligence alone, but from their thoughtful integration. In this new era of discovery, the question isn’t whether to embrace AI in science, but how to do so in ways that amplify human creativity, accelerate beneficial discoveries, and unlock mysteries that have long seemed beyond our reach.