Artificial intelligence and machine learning
Artificial intelligence and machine learning develop algorithms that learn patterns from data to classify, predict or generate outputs. They are increasingly used for hazard detection, forecasting and decision support, and are themselves a source of risk. Artificial Intelligence and Autonomous Systems stewards this page for the Nexus Consortium.
- 01In briefThe short answer
- 02The methodWhat artificial intelligence and machine learning covers
- 03Reference practiceMachine learning in risk work is governed by AI risk frameworks and reporting standards
- 04In systemic risk workFast emulators, fragile extrapolation
- 05Applied inPlatforms that apply this method
- 06StewardshipWho maintains this page
- 07Key termsTerms used on this page
GCRI · Institution · service level R0
The short answer.
Artificial intelligence and machine learning develop algorithms that learn patterns from data to classify, predict or generate outputs. Machine learning can scale detection and forecasting across interconnected systems. In the Nexus Consortium, Artificial Intelligence and Autonomous Systems stewards its canonical page, and 11 other platforms apply it, including Nexus Labs and Nexus Foundry.
Digital demand and connectivity in Southeast Asia.
Urban demand centres and the hubs that anchor digital and data infrastructure, shown for Southeast Asia as an example geography for this subject.
What artificial intelligence and machine learning covers.
- 01
Core methods
supervised learning; deep neural networks; reinforcement learning; representation learning; probabilistic modelling; model interpretation.
- 02
Data and instruments
labelled training datasets; sensor and image archives; text corpora; simulation outputs; benchmark suites.
- 03
Validation
Models are evaluated on held-out data that respects spatial, temporal and group structure. Robustness, calibration and fairness are tested alongside accuracy.
- 04
Limits
Performance degrades when deployment conditions differ from training data. Models can encode biases and fail without clear warning on rare events.
- 05
Use in risk decisions
Machine learning can scale detection and forecasting across interconnected systems. Its own failure modes must be weighed when automated outputs inform consequential decisions.
Machine learning in risk work is governed by AI risk frameworks and reporting standards.
| Reference | What it sets |
|---|---|
| NIST AI RMF | Risk management functions for AI systems |
| ISO/IEC 42001 | AI management systems |
| TRIPOD (EQUATOR) | Reporting of prediction models, including machine-learning extensions |
Fast emulators, fragile extrapolation.
Machine learning speeds up risk analysis: emulators of expensive physical models, classification of imagery, forecasting from large datasets. Its weakness is extrapolation: a model trained on the historical climate or historical markets may fail exactly in the unprecedented conditions that systemic risk is about.
- Test performance out of distribution, not only on held-out samples from the same period.
- Report training data provenance and licence.
- Keep a physical or mechanistic check where one exists.
- Applications on AI and Autonomous Systems and Nexus Labs.
Platforms that apply this method.
- Nexus LabsInstitutionGCRI
- Nexus FoundryInstitutionGCRI
- Artificial Intelligence and Autonomous SystemsInstitutionGCRI
- Future of Work and SkillsInstitutionGCRI
- Innovation NexusInstitutionGRF
- AI and Digital GovernanceInstitutionGRF
- Information Integrity and Public TrustInstitutionGRF
- Insurance NexusInstitutionGRA
- Banking NexusInstitutionGRA
- Fintech NexusInstitutionGRA
- AI and Financial Model RiskInstitutionGRA
- Financial Crime and Fraud RiskInstitutionGRA
Who maintains this page.
Artificial Intelligence and Autonomous Systems stewards artificial intelligence and machine learning: it keeps this description current and reviews how the method is applied across the platforms.
Terms used on this page.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Temporal validation | Testing a model on data from a later period than the data it was trained on, the closest check to how it will perform in use. |
| Calibration | The agreement between a model's predicted probabilities and the frequencies actually observed. |
| Explainability | The degree to which the reasons for a model's output can be understood and stated to the people it affects. |
| Transaction monitoring | Automated screening of transactions for signs of fraud or money laundering. |
Where this page's references lead.
The bodies and documents this page is read against. Status and editions change; each link goes to the issuing body.
Where this page leads across the Consortium.
- Machine learning in laboratory scienceGCRIApplied in Nexus Labs.
- AI and machine learning at Nexus FoundryGCRIApplied in Nexus Foundry.
- Machine learning behaviourGCRIApplied in Artificial Intelligence and Autonomous Systems.
- AI for labour market analysisGCRIApplied in Future of Work and Skills.
- AI as subject and tool in innovation governanceGRFApplied in Innovation Nexus.
- Artificial intelligence and machine learning for AI and digital governanceGRFApplied in AI and Digital Governance.
- Artificial intelligence and machine learning for information integrity and public trustGRFApplied in Information Integrity and Public Trust.
- Machine learning in insuranceGRAApplied in Insurance Nexus.
- Machine learning in bankingGRAApplied in Banking Nexus.
- Machine learning in fintechGRAApplied in Fintech Nexus.
- Artificial intelligence and machine learning for AI and financial model riskGRAApplied in AI and Financial Model Risk.
- Artificial intelligence and machine learning for financial crime and fraud riskGRAApplied in Financial Crime and Fraud Risk.
Questions and answers.
Why does one institution host every method page?
How do I suggest a correction to a method page?
Around GCRI.
Start with an enquiry.
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.