Scope
Aims and scope
Research on artificial intelligence and the systems built from it: methods, infrastructure, evaluation, safety, governance and domain evidence.
Composition
Technology, economics and law, read together
Technology, economics and law are read together when a system in service answers to all three.
- Technology
- the model, the data, the architecture and the engineering. What was built, how it was trained or composed, and what it measurably does.
- Economics and organisation
- the cost of the system, the process it changes, the organisation that has to operate it, and the evidence of the effect it produced.
- Law and regulation
- the rules the system answers to. Supervisory expectation, liability, data protection, auditability, and what has to be demonstrable to an authority.
Subject areas
Work on the systems themselves
The areas the journal covers, revised as the field moves.
- Foundations and methods
- learning theory, model architectures, algorithms, reasoning and representation
- Systems and infrastructure
- training and inference at scale, compute, data pipelines, deployment and operation
- Building with these systems
- software engineering practice, agents and tool use, retrieval, orchestration, and what it takes to keep a system correct in production
- Evaluation and measurement
- benchmarks, reproducibility, statistical practice and the interpretation of results
- Safety, robustness and security
- alignment, failure modes, adversarial behaviour, assurance and verification
- Governance, standards and accountability
- auditability, compliance architecture, assurance regimes, standards and identity
- Society, economics and ethics
- labour, distribution, institutions, environmental cost and public interest
- Human and machine interaction
- interfaces, autonomy, delegation, oversight and trust
Domains
Where these systems meet a field
Domains supply the data, regulatory setting and consequences that make system claims testable.
- Finance and supervision
- banking, insurance, capital markets, valuation, financial crime and supervisory practice
- Audit, assurance and risk
- internal control, model risk, and evidence an auditor can act on
- Law and legal practice
- legal technology, contracting, notarial and judicial process, and compliance architecture
- The public sector
- administration, fiscal policy and public finance, procurement and the delivery of public services
- Health and clinical practice
- diagnosis, care pathways, clinical evidence and patient safety
- Industry and engineering
- manufacturing, mechatronics, energy, logistics and the industrial control estate
- Enterprise architecture
- multi-agent systems, protocol design, ledgers, and running these systems across an organisation
- Work and organisation
- task composition, human oversight, leadership, skills and the design of roles around a system
- Education and assessment
- curriculum, learning design, credentialing and the measurement of what was learned
- The sciences and research practice
- artificial intelligence in the conduct of research, in methodology, and in publishing and peer review
Boundaries
Where the boundary runs
A subject belongs here when it concerns artificial intelligence or a system built from it, and carries evidence a reader can weigh.
- Capability claims
- Capability claims are in scope when the measurement and conditions are reported together. The article states what was measured, under which operating conditions and which evidence supports the claim.
- Adjacent fields
- Statistics, control, operations research, law and the social sciences are in scope wherever they bear on artificial intelligence or on a system built from it, and such work is read on exactly the terms applied to work from inside the field.
- Argument and position
- Arguments without new primary evidence are in scope as Notes when they make a single point the field can act on and situate that point in the literature.
- Surveys and reviews
- A survey is in scope where it contributes a stated search, explicit inclusion criteria and a synthesis that changes how the field reads its own literature.
Particular interest
Artificial intelligence, energy and sustainability
The journal seeks work that reports compute and energy alongside measured performance, including results about the energy systems to which models are applied.
- Efficiency as a result
- What a result costs to obtain is part of the result. Smaller models, distillation, quantisation, sparsity, retrieval in place of scale, better scheduling and higher hardware utilisation are all findings in their own right, and a method that matches a baseline at a fraction of the compute is reported here as the contribution it is.
- Energy systems
- Artificial intelligence applied to generation, grids and storage: forecasting, dispatch and demand response, the integration of renewable capacity, and efficiency in buildings, industry and transport, with the operating evidence that shows what changed.
- Measured footprint
- Method and measurement for the cost of computation itself: how energy, carbon intensity and water use are attributed to training and inference, how the figures are made comparable across sites and hardware, and where attribution requires evidence beyond a datacentre's own numbers.
- ESG evidence and assurance
- Sustainability reporting as an evidentiary problem: the provenance of the data behind a disclosure, the auditability of a claim across a supply chain, and what regulatory reporting regimes require a company to be able to demonstrate.
Readership
Who the journal is written for
Researchers, engineers, and specialists in business and law working on artificial intelligence and the systems it is built into.
- Researchers and engineers
- the people building these systems and measuring what they do, in universities, in laboratories and in industry.
- Practitioners in regulated fields
- the people in finance, audit, law, health, industry and public administration who put a system into service and answer for what it does there.
- Boards, supervisors and policy specialists
- those who set the rules a system runs under, or who have to judge whether an organisation's use of one is defensible.
Author information and submission route are on the authors page.