Digital Minds in 2025: A Year in Review
AI consciousness enters the public discourse, Anthropic leads by example, and the field begins to grow.
December 18, 2025Cambridge Digital Minds
Helping society prepare for the possibility of AI systems with minds
Explore Our WorkOur Work
Based at the University of Cambridge, we conduct interdisciplinary research on the societal, policy, and governance implications of digital minds, with an emphasis on anticipating future debates and decision points.
Research
Studying public, expert, and policymaker beliefs on AI consciousness and welfare through surveys, forecasting, scenario planning, and citizens' assemblies.
Field Building
Strengthening the digital minds and AI welfare field through fellowships, training programs, conferences, and newsletters anchored in academic rigor.
Expert Coordination
Engaging with policymakers, industry, and the public to reduce confusion and polarization, and shape responsible standards before high-stakes conflicts emerge.
Research
Our recent publications on digital minds, AI consciousness, and AI welfare.
Chapter
Preprint
Report
Journal
AI Consciousness Will Divide Society
Argues that AI consciousness will become a major source of societal division. Analyzes how stakeholders—researchers, policymakers, industry, and the public—will form conflicting views on AI moral status. Examines historical parallels and proposes strategies for reducing polarization before high-stakes conflicts emerge.
The ML Community Must Prepare for AI Consciousness, Perceived or Real
Addresses the ML community directly, arguing that AI consciousness—whether genuine or perceived—will create challenges requiring advance preparation. Outlines concrete steps researchers can take now, including evaluation frameworks and ethical guidelines. Emphasizes that waiting for scientific consensus is not viable given the pace of AI development.
Subjective Experience in AI Systems: What Do AI Researchers and the Public Believe?
Survey of 582 AI researchers and 838 US public participants on beliefs about AI developing subjective experience. Both groups estimate meaningful probabilities this century (25-30% by 2034, 60-70% by 2100), but disagree on governance and whether protective measures should be implemented now.
What Will Society Think About AI Consciousness? Lessons from the Animal Case
Examines how debates over animal consciousness can inform predictions about societal responses to AI consciousness. Analyzes patterns of public attitude formation, expert disagreement, and policy development, drawing lessons for how society might respond to sophisticated AI systems.
Featured
Guides, reports, podcasts, and more from our work on digital minds.
Guide
When AI Seems Conscious
A practical guide for people who wonder or are concerned whether AI chatbots are conscious. More and more people report believing their AI is conscious based on personal conversations. We created this public resource to help.
The guide makes two key points: Today's AIs are probably not conscious, but we cannot be certain. Current AIs are highly skilled at appearing conscious, and humans are prone to projecting agency onto them. But it's still important to take AI consciousness seriously—future systems could be conscious, and that possibility demands preparation.
Visit WhenAISeemsConscious.org →
Report
Futures with Digital Minds: Expert Forecasts in 2025
We surveyed 67 experts about whether, when, and how digital minds might be created. Key findings: It's very likely that digital minds are possible in principle, with a 50% median estimate that they will be created by 2050.
Conditional on digital minds arriving by 2040, their collective welfare capacity could exceed humanity's within a decade. There's nothing approaching consensus on whether their welfare will be positive or negative—humanity's poor track record with vulnerable groups is reason for concern.
Read the full report →Fellowship Programme
A selective residential fellowship for early- and mid-career researchers working on digital minds, AI consciousness, and AI welfare.
Digital Minds Fellowship
A 5-day intensive residential programme hosted at Cambridge University. The fellowship enables deep, cross-disciplinary engagement across philosophy, social science, technical research, policy, and governance, fostering shared norms, judgment, and coordination capacity.
Programme Highlights
- Structured teaching and discussion sessions
- One-on-one mentoring with senior researchers
- Independent and small-group project development
- Career and field-building sessions
- Direct participation in the Expert Workshop
Expert Workshop
A two-day, output-oriented workshop bringing together fellows, mentors, and invited experts for structured research on digital minds.
Day 1
AI Welfare Intervention Assessment
Systematic evaluation of candidate interventions across effectiveness, tractability, risks, and interaction with AI safety.
Research PaperDay 2
Expert Panel & Consensus
Structured elicitation of views on timelines, key uncertainties, and concrete policy recommendations.
Policy DocumentAlt
Long-term Scenario Planning
Developing scenarios through ~2100 and generating novel strategic intervention ideas.
Strategic ReportFellowship alumni automatically attend.
Learn MoreOnline Course
Build foundational understanding of digital minds, AI consciousness, and AI welfare through our facilitated online programme.
Launching 2026
Introduction to Digital Minds
An 8-week facilitated course designed to equip participants with shared vocabulary and conceptual grounding. Learn to reason well under uncertainty and engage responsibly with emerging debates on AI consciousness and moral status.
8 Weeks
Facilitated learning
Project Phase
Optional 4-week extension
Fellowship Track
Pathway to in-person programme
About Us
Cambridge Digital Minds is a research initiative at the University of Cambridge focused on societal preparedness for digital minds.
Our Mission
Help society prepare to make accurate, ethical, and well-reasoned decisions around digital minds, including questions of consciousness, welfare, and moral status, even under conditions of deep uncertainty and disagreement.
Our Approach
We focus on improving societal decision-making under uncertainty by anticipating societal dynamics, clarifying areas of expert agreement and disagreement, developing responsible frameworks and standards, and creating credible institutions that can inform policy and public debate.
Our Team
- Lucius Caviola — Director
- Will Millership — Director of Operations
- Henry Shevlin — Associate Director
- Bridget Harris — Online Course Lead
- Pooja Khatri — Online Course Lead
- Ali Ladak — Postdoctoral Researcher
- Brad Saad — Senior Researcher (Affiliate)
3
Core Activities
2026
Fellowship Launch
50+
Workshop Participants
8
Week Online Course