Why it matters
AI-for-science is where frontier-model capability claims get stress-tested by people who run actual labs — bio, chem, materials, climate. The critique-heavy format surfaces what works versus what's demo-ware.
The tokenmaxxing angle
Scientific workloads — literature triage, hypothesis generation, lab automation loops — are token-hungry and budget-sensitive; grant-funded labs feel inference pricing more sharply than startups burning VC credits.
From the organizers
Agenda: doors 5:30 PM, networking and light dinner to 6:30, presentations 6:30-8:00, networking to 8:30. All talks listed as TBA; capacity limited, host approval required; 'No ML background required.'