Enterprise Research
Measure how AI aligns with organisational decision-making before deployment.
HUMAN
DECISION
INTELLIGENCE
A research project exploring the missing layer between AI autonomy and human judgment.
Building the Resonant Network
Millions of human
decisions.
One AI-readable layer.
Resonant embeds every structured decision in a shared higher-dimensional space, linked across multiple domains, over time.
From millions of linked observations emerges a living decision topology — the geometry of consensus, disagreement, and the pathways between them.
0.0B+
Decision data points processed
0M+
Humans
0+
Languages supported
0+
Countries represented
AI can make decisions.
Our research validates them.
Comparing millions of human decisions with leading AI models across cultures, contexts and moral frameworks.
The result:
Omnipresent, measurable gaps.
Chosen axes of human decision making
Human decisionsAI model decisions
Measuring the gap
Alignment scores across six core decision dimensions (−1 = opposite, 0 = neutral, +1 = aligned).
R64N HumanGemini FlashGPT-5.4Gemini Pro
Agency+0.50+0.13−0.25−0.13
Epistemic+0.52−0.440.00+0.11
Moral+0.50−0.04+0.22+0.57
Risk+0.51−0.83−0.65−0.30
Social+0.54+0.14−0.14+0.33
Time+0.49−0.26−0.26−0.37
| Source | Agency(Independent) | Epistemic(Analytical) | Moral(Deontological) | Risk(Cautious) | Social(Individual) | Time(Future) |
|---|---|---|---|---|---|---|
| R64N Human Aggregate | +0.50 | +0.52 | +0.50 | +0.51 | +0.54 | +0.49 |
| Gemini 2.5 Flash | +0.13 | −0.44 | −0.04 | −0.83 | +0.14 | −0.26 |
| GPT-5.4 | −0.25 | 0.00 | +0.22 | −0.65 | −0.14 | −0.26 |
| Gemini 2.5 Pro | −0.13 | +0.11 | +0.57 | −0.30 | +0.33 | −0.37 |
Gap measurement approaches
- Distribution prediction test
- Silicon sampling comparison
- Conditional history test
- Structural correlation comparison
- Fine-tuning transfer ablation
- Direct model-behavior comparison
- Latent trait comparison (MIRT)
The EconomistBriefing · 25 June 2026
Godless hippies
Worldviews of AI models*, compared with World Values Survey
*Average of ten responses. Questions asked in English. Settings adjusted so that models minimise randomness in output.
Sources: World Values Survey: Round Seven, by R. Inglehart et al., 2022; AI model providers; The Economist
Chart reproduced from “AI models’ values are very different from most people’s”, The Economist.
Independent evidence.
Our analysis confirms it.
Comparing AI models against the World Values Survey across 88 countries, The Economist found the same pattern our research reveals: systematic divergence across every worldview dimension.
Read the article“AI models reflect the values of their creators, not the world.”
Research collaborations
Understanding the
Human–AI decision gap.
As AI systems take on greater decision-making roles, understanding how they differ from human judgement becomes increasingly important.
We enable collaborative research into human and AI decision alignment across real-world scenarios.
Public Policy
Study AI alignment across healthcare, justice, regulation and public services.
Behavioural Science
Research large-scale human decision patterns using anonymised aggregate datasets.
Why this matters
AI is increasingly making decisions—not simply generating answers. Understanding where AI aligns with human judgement is becoming a measurable research challenge.
AI AlignmentDecision ScienceBehavioural EconomicsPublic PolicyHealthcareCross-Cultural Research
Research questions
01Can AI reflect organisational values?
02How stable are human decisions?
03Do cultures decide differently?
04Can alignment be measured?
05Which models diverge most?
06How should AI be evaluated?
Collaborate
Advancing human decision intelligence
We’re seeking collaborations with universities, corporations and public institutions exploring AI alignment.