Artificial Intelligence and Autonomous Systems
The Global Centre for Risk and Innovation (GCRI)
Maps of Artificial Intelligence and Autonomous Systems are published here once their data rights are cleared and they have been reviewed. Each shows its time and space reference, its sources and its uncertainty.
Public service pilot · Layers 6 · Scale 1:244,000
schematic · placeholder geography
| Models in registry | OFFICIAL | On |
| Evaluations passed… | ASSESSED | On |
| Agent actions… | OBSERVED | On |
| Compute per training… | REPORTED | Off |
| Administrative… | OFFICIAL | Off |
Artificial Intelligence and Autonomous Systems · Map workspaceMap workspace, in outline: domain structure with placeholder names and illustrative values, not a record.
Artificial Intelligence and Autonomous Systems released maps are reviewed map views of artificial intelligence, published once their data rights are cleared, each with its coordinate reference, date, sources and uncertainty. Layers draw on sources such as Stanford HELM (Holistic Evaluation of Language Models). A map shows released, rights-cleared evidence, not a live feed or an official warning.
Fig.A programme from scoping to the owner's decision. Design, not a record.
For released maps, in GCRI's spatial-intelligence reading, the map opens on night lights, true-colour imagery and population density with true colour from NASA GIBS, framed on World; 3 further NASA layers wait in the list.
The interactive map needs JavaScript; the layer list and sources are below.
Layers 3
Energy now, for this page
Keyboard, with the map focused
Hold Ctrl or ⌘ and scroll to zoom
Latest available
NASA GIBS / EOSDIS · ASTER GDEM (METI, NASA) · Blue Marble (NASA) · © OpenStreetMap contributors (ODbL) · Natural Earth
Loading the latest NASA imagery.
Drag to move. Ctrl or ⌘ and scroll, pinch or the buttons to zoom; hover for the values under the pointer; reset returns to the region.
For released maps, in GCRI's spatial-intelligence reading, the map opens on night lights, true-colour imagery and population density with true colour from NASA GIBS, framed on World; 3 further NASA layers wait in the list.Satellite and model products as NASA publishes them, each with its own date; indicators, not assessments of any place; claim ceiling INTEGRATED.
Layers and dates
| Indicator | Unit | Cadence | Evidence label | Topic |
|---|---|---|---|---|
| Benchmark score with uncertainty | score with confidence interval | per evaluation | Reported | Model evaluation |
| Contamination checks reported | percent of evaluations reporting checks | per evaluation | Declared | Model evaluation |
| Red-team findings by severity | count by severity class | per exercise | Reported | Model evaluation |
| Performance gap between subgroups | difference in error rates | per evaluation | Observed | Model evaluation |
| Post-deployment incident reports | count by type | monthly | Reported | Model evaluation |
| Calibration error | expected calibration error | per evaluation | Observed | Model evaluation |
| Task success rate on agent benchmarks | percent of tasks completed correctly | per evaluation | Reported | Agentic systems |
| Actions requiring human approval | percent of consequential actions | per deployment | Declared | Agentic systems |
| Prompt injection success rate | percent of attempts succeeding | per evaluation | Observed | Agentic systems |
| Agent incidents | count by severity | monthly | Reported | Agentic systems |
| Actions logged and attributable | percent of agent actions | per deployment | Observed | Agentic systems |
| Mean task length before human intervention | steps or minutes | per evaluation | Observed | Agentic systems |
| Vulnerabilities reported in AI frameworks and libraries | count of CVE entries | monthly | Reported | Ai cyber security |
| Adversarial robustness | accuracy under a stated attack | per evaluation | Observed | Ai cyber security |
| Models with signed provenance | percent of deployed models | per inventory | Declared | Ai cyber security |
| AI-related security incidents | count by technique | annual | Reported | Ai cyber security |
| Time to detect AI system compromise | hours | per incident | Observed | Ai cyber security |
| Organizations with AI security policies | percent of surveyed organizations | per survey | Reported | Ai cyber security |
| Studies sharing code and data | percent of published ML studies | per review | Observed | Ai in science |
| Leakage issues identified in reviews | count by type | per review | Reported | Ai in science |
| Out-of-distribution performance | error on held-out regimes | per evaluation | Observed | Ai in science |
| Emulator error against reference simulation | error metric | per evaluation | Observed | Ai in science |
| Journals with AI use policies | count | annual | Declared | Ai in science |
| Compute used for scientific model training | floating point operations | per model | Estimated | Ai in science |
| Public algorithm registers | count of systems registered | per update | Official | Ai and public services |
| Algorithmic impact assessments completed | percent of deployed systems | annual | Declared | Ai and public services |
| Error rate by population group | difference in error rates | per audit | Observed | Ai and public services |
| Decisions appealed and overturned | count and percent | annual | Official | Ai and public services |
| Systems with human review of adverse decisions | percent of systems | per review | Declared | Ai and public services |
| Audit findings on public sector algorithms | count by severity | per audit cycle | Reported | Ai and public services |
| Crashes involving automated driving systems | count by severity | monthly | Official | Autonomous physical systems |
| Disengagements or interventions | count per distance or hour | annual | Reported | Autonomous physical systems |
| Distance driven autonomously | kilometres | annual | Declared | Autonomous physical systems |
| Industrial robot injuries | count | annual | Official | Autonomous physical systems |
| Operating domain exceedances | count of events outside design domain | per deployment | Observed | Autonomous physical systems |
| Drone incidents near aircraft | count | annual | Reported | Autonomous physical systems |
| Data centre electricity consumption | terawatt hours per year | annual | Estimated | Compute and energy |
| Power usage effectiveness | ratio of facility to IT energy | annual | Declared | Compute and energy |
| Water usage effectiveness | litres per kilowatt hour | annual | Declared | Compute and energy |
| Grid connection requests from data centres | megawatts requested | quarterly | Official | Compute and energy |
| Training compute of notable models | floating point operations | per model | Estimated | Compute and energy |
| Energy per inference | watt hours per query | per measurement | Estimated | Compute and energy |
| Employment in occupations highly exposed to AI | percent of employment | per assessment | Estimated | Labour and social effects |
| Firms using AI | percent of firms | annual | Reported | Labour and social effects |
| Workers using AI tools at work | percent of workers | per survey | Reported | Labour and social effects |
| AI-related job postings | share of postings | monthly | Observed | Labour and social effects |
| Participation in AI skills training | percent of workforce | annual | Reported | Labour and social effects |
| Wage changes in exposed occupations | percent change | annual | Official | Labour and social effects |
Analysis runs in Nexus Foundry; reviewed releases are listed here.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Coordinate reference system | The definition that ties coordinates to locations on Earth, catalogued with EPSG codes. |
| Modifiable areal unit problem | The change in statistical results that comes from how areas are drawn and aggregated. |
| Zonal statistics | Statistics of raster values summarised within boundaries such as districts. |
| Public-safe release | A recorded decision on the detail at which a layer may be published without exposing sensitive sites or people. |
The bodies and documents this page is read against. Status and editions change; each link goes to the issuing body.
The current service level is R0: information and enquiries. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.