AI in Science
AI research is moving fast - including for scientific discovery. AI in Science brings together leading researchers from Karolinska Institutet (KI), KTH Royal Institute of Technology, and Stockholm University for an accessible, high-energy evening that highlights the breadth of what’s being worked on right now.
This summit offers three distinct perspectives into today’s frontier:
Anandi Hattiangadi
Stockholm University, Department of Psychology
Title:
Artifical General Intelligence: A Philosopher's Manifesto
Astract:
In this talk, Professor Hattiangadi argues that current conceptions of artificial general intelligence (AGI) are misguided. Although the concept of AGI is typically indexed to human intelligence, it is indexed to philosophical accounts of human intelligence that are fundamentally flawed. This distorts AGI timelines and cost-benefit assessments. To fix the problem, Hattiangadi argues, we need to return to philosophical foundations: what makes human intelligence distinctively general? What it would take for a machine to have an intelligence to rival that of a human? Through addressing these questions, Hattiangadi highlights the flaws in the design of existing systems. But she also illuminates a promising path forward, one that requires close collaboration between philosophers and AI experts.
Belén García Pascual
Karolinska Institutet (Neurobiology) & KTH (Mathematics)
Title:
Topological Data Analysis for AI-based Personalised Dementia Treatments
Astract:
Personalised medicine depends on making sense of complex patient data - where traditional approaches can miss structure and nuance. García Pascual presents research that combines AI with advanced mathematical methods to analyze dementia-related clinical data. By identifying meaningful patient subgroups, mapping medication patterns, and extracting clinically relevant insights, this work aims to support more tailored and effective treatment strategies.
André Silva
KTH, Computer Science
Title:
Training Agentic Code Models
Astract:
Code models are evolving from “autocomplete” tools into systems that can reason, plan, and act. Silva’s talk explores the training and improvement of agentic code models-models designed to carry out software engineering tasks more autonomously, including iterative problem-solving and code refinement. The session highlights what it takes to build models that don’t just generate code, but can make progress on complex development workflows.
Arrival: Doors open, food, network, 17:00 - 17:30
Opening: Stockholm AI intro 17:30 - 17:40
Prof. Anandi, The future of AI: Talk 17:40 - 18:10
PhD Belén, Medical AI Impact: Talk 18:15 - 18:35
PhD André, Agentic Code Models: Talk 18:50 - 19:20
Networking: Venue, food, network, 19:20 - 20:00