Historical overview
Formal study traces to Émile Durkheim's collective consciousness (1893), was advanced by Pierre Lévy's 1994 book 'Collective Intelligence', and is now central to network science, prediction-market research, and the study of large open-source projects.
Scientific basis
Collective intelligence depends on diversity, independence, decentralisation, and aggregation mechanisms (Surowiecki, 2004). Information cascades, communication topology, and incentive alignment determine whether a group outperforms its best individual member or collapses into groupthink.
Strengths
- Aggregates noisy individual estimates into accurate group judgments
- Robust to single-point failures
- Scales with population size and connectivity
Limitations
- Vulnerable to information cascades, manipulation, and polarisation
- Coordination overhead grows non-linearly with group size
- Lowest-common-denominator outcomes when diversity is suppressed
Relationship to other intelligence systems
Human Intelligence
The substrate that makes collective intelligence possible.
Artificial Intelligence
Multi-agent AI systems are a synthetic form of collective intelligence.
Future implications
Human–AI hybrid collectives - where humans and language models share a decision graph - are emerging as a distinct organisational form.