Humanity Explorer is a software tool developed by the FHSR team that builds synthetic people and cultures to create virtual societies for research, policy testing and education without involving real individuals. The project is led publicly by anthropologist Dávid-Barrett Tamás, who teaches at Trinity College, Oxford, is a member of the Royal Anthropological Institute and a visiting researcher at the Helsinki Research Centre for Population.
The stated aim is not to predict the future or produce digital copies of real people, but to expand the space in which humans can explore the consequences of decisions. As Dávid-Barrett put it: “We are not a forecasting company; we are a company that supports human thinking.”
The problem the system addresses
Human cognition evolved to understand small groups: people typically track at most a few dozen to a few hundred others with detail. The modern world, however, includes more than eight billion people and highly diverse cultures, so most distant societies are represented in our minds by coarse, often outdated categories. Humanity Explorer seeks to widen that social field of view by enabling comparisons of the same life situations — such as childbearing, love, loneliness, or aging — across multiple cultural variants.
How the synthetic people work
The system constructs synthetic populations: virtual people with their own backgrounds, social environments, attitudes and cultural embeddings. These are not reproductions of real individuals; the user is not present in the model, although the tool can search for synthetic persons with profiles similar to a user.
Current generations of agents do not yet form fully autonomous societies: they do not continuously birth, age and die within the model. Instead, the platform resembles a very large, dynamic social snapshot: characters are embedded in villages, cities, regions and cultures and can perceive and respond to their surroundings. One core engineering challenge has been making agents both sufficiently unique and internally consistent.
Simulating emotions is harder, because human behaviour is not determined solely by rational calculation. People often provide rational post hoc accounts for choices driven by strong emotions; that discrepancy matters especially for areas such as partner choice or fertility, where survey answers can diverge from actual behaviour.
Intended uses
The developers describe several practical and research uses, while stressing the need for validation:
- Trialling policy or institutional changes: governments or municipalities can compare how different social groups react to multiple variants of a measure before implementation.
- Pre-screening research: running hypotheses and checking model stability before committing to costly, logistically challenging fieldwork.
- Creating focus groups impossible to convene in reality: for example, surveying people from a Panamanian rainforest community, a Namibian desert community, a highland village and an Amazonian group together to see differing perspectives on tourism or environmental interventions.
- Reaching groups underrepresented in traditional surveys: very wealthy individuals, marginalised communities, or people who typically avoid polls and focus groups.
- Education: as a tool for schools to develop cultural intuition, helping pupils recognise that different cultures operate from different norms, stories and value systems.
Dávid-Barrett recounted a modelling exercise on family policy where modest child allowances had negligible effects on behaviour, while dramatically larger payments produced measurable change. He emphasises that this does not mean the model prescribes exact policy levels, but that it allows hypothesis testing in an experimental environment.
Validation and ethical considerations
FHSR’s approach differs from operations such as Cambridge Analytica: whereas Cambridge Analytica built voter profiles from large numbers of real-person data points, Humanity Explorer first constructs artificial people and then calibrates the population to empirical data. The project leaders insist on continuous validation for each use case. They also say that monopolisation of the technology would be dangerous and therefore aim to broaden access.
A principal intellectual risk, according to Dávid-Barrett, is overtrust in the model. A synthetic population’s response does not by itself determine what policies should be pursued; the key question often becomes what institutional arrangements and coordination mechanisms would change behaviour among many actors.
Technical status and next steps
At the time of the interview, the system supported roughly 80 languages, which the team says could cover the majority of the world’s population in their native tongues. The next major development milestone would be turning the snapshot-like populations into truly generative societies where synthetic people are born, form relationships, influence each other and are replaced across generations. That would shift the platform from mapping a social state to modelling social processes.
Dávid-Barrett is cautious about turning the tool into a digital oracle. He compares the platform to scientific instruments like microscopes or maps: they let us see more, but do not tell us what to conclude or what route to take. Humanity Explorer’s promise is therefore modest but practical: to reveal more of the eight billion-people world that is invisible to individuals while providing a controlled space to ask questions that would be too expensive, slow or dangerous to test in reality.
Why it matters
If validated and used responsibly, the platform could serve as a laboratory for family policy experiments, a measurement tool for opinion research, a focus-group generator for otherwise unreachable participants, or an educational aid for cultural literacy. Whether it will genuinely foster better mutual understanding across cultures remains uncertain; the developers consider it a possible outcome but make no guarantee.
