Connecting Cambridge in AI

Revolution

The Future of Professions

The Innovation Gap

New Productivity Paradox

• Current productivity flywheel relies on measurement to translate innovation into productivity.
• Without measurement how does the wheel spin?

What Do People Want?

Public Research

Public Research

Public Research

Public Research

AI in Education: Public Perspectives

• Support for reducing teacher workload.
• Concerns about screen time and support for human interaction.

AI in Education: Public Perspectives

“Education isn’t just about learning, it’s about preparing children for life, and you don’t do all of that in front of a screen.”

Public Participant, Cambridge ai@cam and Hopkins Van Mil (2024) pg 18

AI in Education: Public Perspectives

“Kids with ADHD or autism might prefer to interact with an iPad than they would a person, it could lighten the load for them.”

Public Participant, Liverpool ai@cam and Hopkins Van Mil (2024) pg 17

The Attention Reinvestment Cycle

The Attention Reinvestment Cycle

• Traditional innovation flywheel: R&D → Innovation → Productivity → Economic Surplus → R&D
• Problem: Misalignment with societal needs in healthcare, education, social care

A New Model for Innovation

• Attention Reinvestment Cycle liberates human attention
• Freed attention reinvested in knowledge networks
• Creates value through improved human services

Putting It Into Practice: Connecting Cambridge

Why Connecting Cambridge?

• Universities organised in silos
• Real problems don’t respect boundaries
• Need to connect expertise across disciplines

Cambridge’s Unique Position

• Extraordinary breadth: 100+ departments, 31 colleges
• Deep expertise across AI, education, health, policy, ethics
• Challenge: connecting capability to needs

Two Complementary Initiatives

• ai@cam: University-wide mission across 6 schools
• Accelerate Science: Science supporting-focused programme
• Complementary approaches to connecting expertise

From Connection to Impact

• Connect researchers across boundaries
• Share tools, infrastructure, knowledge
• Build capacity across the ecosystem

ai@cam: The University Mission

• University flagship programme (£5M, 2022)
• Works across 6 schools, 30+ departments
• Bridges research, policy, and practice

Accelerate Science: Building Research Capacity

• Programme for scientific machine learning
• Knowledge networks and peer learning
• Builds capability across disciplines

Example: Pico Language Models

• Modular toolkit for small language models (1M-1B parameters)
• Research-ready framework for systematic experimentation
• Open science approach to understanding learning dynamics

• pico-train: Easy model training with minimal configuration
• pico-analyze: Deep analysis of learning dynamics
• Value reinvested through open tools and shared knowledge

Work by Richard Diehl Martinez and Paula Buttery et al.

Example: CMBAgents

• Multi-agent systems for automating scientific discovery

• Open-source research backend (cmbagent)

• End-to-end research system (Denario)

• Autonomous research tools free researcher attention

• Value reinvested through open-source sharing

• Building capacity for AI-assisted research

Work by Licong Xu and Boris Boillet et al.

The Connecting Pattern

• Connect researchers across disciplines
• Connect tools to users
• Connect innovations to needs

The Path Forward

Innovation Economy Conclusion

• Interact directly with micro-demand

• Release quality attention

• Reinvest human capital in more innovation

• Measure value in freed human attention

• Reinvest in knowledge networks

• Build capacity for addressing real needs