Trades and AI glossary
Updated · 13 sources
Research on AI and work uses words that sound interchangeable but measure different things. Exposure is about tasks a technology could affect; adoption is about tools actually used; displacement is about work or employment changing. These definitions, with the route terms beside them, help you read a source before drawing conclusions about a career or paying for training.
Research terms
- AI exposure
- The extent to which a study judges an occupation's tasks or required abilities could be affected by AI. It says nothing on its own about whether a tool will be adopted, a worker will be helped, or a job will disappear. Department for Education (Unit for Future Skills).
- AI complementarity
- The possibility that AI helps someone perform part of their work while their judgement and other tasks remain important. A job can be highly exposed to AI and still have scope for tools to complement its workers. Department for Science, Innovation and Technology and the AI Security Institute.
- AIOE score
- An occupational exposure score built by relating AI capabilities to abilities used in an occupation group. The Department for Education uses it for comparisons, while warning that a relative score is not a precise measure of employment impact. Department for Education (Unit for Future Skills).
- Augmentation
- A change in which a tool supports a person doing a task, such as drafting information for a worker to check. It differs from substitution, and an exposure index alone does not establish which outcome will follow. Department for Education (Unit for Future Skills).
- Automation and substitution
- Automation means technology performs some work that people previously did. Substitution is the narrower possibility that it takes over a task or role. A study of what is technically possible is not proof that an employer will make that change. Department for Education (Unit for Future Skills).
- Displacement scenario
- A model of what could happen to work under stated assumptions about technology and adoption. Scenarios help compare possibilities; they should not be reported as observations or reliable predictions for a particular worker. Institute for Public Policy Research.
- Task-based analysis
- A way of examining the separate activities inside a job rather than treating its title as a single indivisible unit. It can reveal that information tasks and physical site tasks face different kinds of change. Department for Education (Unit for Future Skills).
- Generative AI
- Software that can create text, images or other outputs in response to instructions. Its ability to draft or summarise material can affect parts of a job, while saying little about the physical work carried out on site. Department for Science, Innovation and Technology and the AI Security Institute.
- Large language model
- A system trained to process and generate language. The Department for Education reports a separate exposure ranking for language modelling, which can differ from its broader measure covering other AI applications. Department for Education (Unit for Future Skills).
- AI agent
- Software that can take a sequence of steps towards a goal using AI tools. The name describes a way of arranging software, not proof that it can perform every task in an occupation or carry legal responsibility for work. Department for Science, Innovation and Technology and the AI Security Institute.
- Exposure versus adoption
- Exposure estimates where AI could affect tasks; adoption records whether people or organisations actually use a tool. A high exposure estimate can coexist with limited use, and neither measure alone counts jobs lost. Office for National Statistics.
- AI applicability
- A judgement about where an AI capability might be useful for a work activity. Microsoft's US preprint measured it from chatbot conversations. It is narrower than a forecast of employment, because usefulness does not show whether a business will buy the tool or change staffing. Microsoft Research (arXiv preprint).
- Occupation, job and task
- An occupation is a statistical group, a job is someone's actual role, and a task is one activity within it. Research often assigns a group score, so two people in the same occupation can have quite different day-to-day work. Department for Education (Unit for Future Skills).
- SOC code
- A Standard Occupational Classification identifier used to group related work in official statistics. The Department for Education's exposure ranking uses 365 groups under SOC 2010; the code is a category, not a personal score.[1] Department for Education (Unit for Future Skills).
- Early-career effect
- A possible difference in how new entrants and experienced workers are affected by a technology. Evidence about one age or experience group should be read with its country, sample and time period, rather than generalised to everyone. Stanford Digital Economy Lab.
- Knowledge work
- Work centred on information, language, analysis or decisions. Some studies find these activities more exposed to current text tools than many physical tasks, but any one role can contain both kinds of work. Department for Education (Unit for Future Skills).
- O*NET
- A US database describing the abilities and activities each occupation involves. The Department for Education's ranking maps it onto UK occupation groups, because there is no equivalent UK database, so a UK rank inherits US job descriptions. Department for Education (Unit for Future Skills).
- Preprint
- A research paper shared publicly before, or instead of, formal peer review. arXiv says the material it hosts is not peer-reviewed by arXiv, so we label preprints, such as Microsoft's study of chatbot use, wherever we cite them. arXiv.
- Official statistics in development
- A label for new or changing official statistics. The ONS figures on business use of AI carry it, and the ONS says results from newly added survey questions should be read as early indications. Office for National Statistics.
- Workforce forecast
- An estimate of how many extra workers an industry will need, given assumptions about future output and people leaving. CITB's construction outlook is one; it measures recruitment needed, not vacancies open today. Construction Industry Training Board (CITB).
- Apprenticeship
- A paid job combined with training and study, with at least a fifth of normal working hours set aside for learning in England. Scotland, Wales and Northern Ireland run their own apprenticeship systems. Department for Education (GOV.UK guide).
- Free Courses for Jobs
- An England scheme that pays course fees for some adult qualifications, including some Level 2 courses in construction, for people who meet its age and earnings conditions. A funded place is not a job offer. Department for Work and Pensions.
- Advanced Learner Loan
- A repayable loan for eligible college or training courses in England, repaid once income passes a threshold. GOV.UK advises asking the provider about a grant first for Level 3 courses, which does not need repaying. Student Loans Company (GOV.UK guide).
- “AI-proof”
- A search phrase people use when asking about their future work, not a research term: the government review cited here does not use it. We use it only as a question, because research compares exposure and possible changes but cannot promise that any occupation will stay the same. Department for Science, Innovation and Technology and the AI Security Institute.
Electrical route terms: notifiable work, ECS card and AM2.
What the evidence says
A score needs a method
The Department for Education's AIOE method compares the abilities required in occupation groups with AI capabilities. The score is relative.
Source: Department for Education (Unit for Future Skills) · 28 November 2023 · UK
Exposure has limits
A government review describes occupational exposure indices as indicative rather than precise. Different studies can measure different capabilities.
Source: Department for Science, Innovation and Technology and the AI Security Institute · 28 January 2026 · UK
Adoption is a separate measure
The ONS asks businesses whether they use AI at all. Its figures carry the label official statistics in development, so early results may change as the questions settle.
Source: Office for National Statistics · 20 July 2026 · UK
A route is separate
The National Careers Service describes actual electrician duties and training routes. A research label does not replace exploring the work.
Source: National Careers Service · undated · England
What it doesn’t tell you
- A definition can clarify a study without making its result transferable to every country, job or employer.
- A study of software tools does not automatically cover physical robotics or all work done on a site.
- Qualification and registration terms have legal or scheme-specific meanings; a lesson completion is neither.
- Funding and apprenticeship terms here describe England. Scotland, Wales and Northern Ireland use their own schemes and names.
DfE caveat: The analysis measures the exposure of jobs to AI, rather than distinguishing whether a job will be augmented (aided) or replaced (substituted) by AI.
Translate the claim before judging it
Rewrite an AI headline using the terms in this glossary: what technology, which tasks, whose jobs, what country and what timescale? If exposure has become redundancy, or a scenario has become a forecast, go back to the study. The distinction matters more than a dramatic label.
Read the source: Department for Science, Innovation and Technology and the AI Security Institute
If you’re weighing up a trade
Use these terms to check what a headline really measures, then try a short electrical workshop if you want to explore the work itself.
Doing it yourself? Read this first
Some electrical work in your own home is notifiable under Part P of the Building Regulations — a new circuit, a consumer-unit change, or work on a circuit close to a bath or shower — and you cannot certify it yourself. There are two main lawful routes: an electrician registered with a competent-person scheme certifies and notifies it, or you tell your local authority building control before you start and pay for the inspection. In England only, an installer who is not scheme-registered can instead book a registered third-party certifier before work starts. Part P covers England and Wales, and Wales notifies more kinds of work than England does; Scotland and Northern Ireland run their own building standards. Every job, notifiable or not, still has to meet BS 7671. We teach you how the work is done and why — finishing a lesson is not a certificate, and it does not make you a competent person under any scheme.
Common questions
- Is AI exposure the same as job loss?
- No. It identifies tasks that may be affected; it does not say whether they will be helped, changed or removed.
- What is a SOC code?
- A Standard Occupational Classification code groups related jobs for statistics. A group does not describe every worker within it.
- Where are electrical regulation terms?
- Use our electrical glossary for notifiable work, the ECS card and the practical assessment.
Sources and checks
1.The impact of AI on UK jobs and training · Department for Education (Unit for Future Skills)
Published: 28 November 2023. Checked: 23 October 2026. Coverage: UK. Kind: exposure index. Licence: Open Government Licence v3.0.
- 365: occupation groups ranked (UK, exposure index). Groups, not individual jobs or workers; exposure is not a job-loss forecast.
Supports: The AIOE score, occupation groups and exposure terms. Conflict: None declared in this source record. Data file
2.Assessment of AI capabilities and the impact on the UK labour market · Department for Science, Innovation and Technology and the AI Security Institute
Published: 28 January 2026. Checked: 23 October 2026. Coverage: UK. Kind: evidence review. Licence: Open Government Licence v3.0.
Supports: Interpretation and limits of exposure measures. Conflict: None declared in this source record.
3.Electrician: job profile · National Careers Service
Published: undated. Checked: 23 October 2026. Coverage: England. Kind: career profile. Licence: Open Government Licence v3.0.
Supports: The distinction between research terms and a real trade route. Conflict: None declared in this source record.
4.Artificial intelligence in UK businesses · Office for National Statistics
Published: 20 July 2026. Checked: 23 October 2026. Coverage: UK. Kind: survey. Licence: Open Government Licence v3.0.
Supports: Adoption and the official statistics label. Conflict: None declared in this source record.
5.Up to 8 million UK jobs at risk from AI unless government acts · Institute for Public Policy Research
Published: 27 March 2024. Checked: 23 October 2026. Coverage: UK. Kind: scenario. Licence: © IPPR; short citation only.
Supports: Displacement scenarios. Conflict: None declared in this source record.
6.Working with AI: Measuring the Applicability of Generative AI to Occupations · Microsoft Research (arXiv preprint)
Published: 10 July 2025. Checked: 23 October 2026. Coverage: US. Kind: preprint. Licence: arXiv author submission; citation only.
Supports: AI applicability. Conflict: Microsoft develops and sells AI products.
7.Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence · Stanford Digital Economy Lab
Published: 12 August 2026. Checked: 23 October 2026. Coverage: US. Kind: working paper. Licence: © authors; short citation only.
Supports: The early-career effect. Conflict: None declared in this source record.
8.About arXiv · arXiv
Published: undated. Checked: 23 October 2026. Coverage: global. Kind: method. Licence: © arXiv; short citation only.
Supports: What a preprint server does and does not review. Conflict: None declared in this source record.
9.Construction Workforce Outlook 2026–2030: United Kingdom · Construction Industry Training Board (CITB)
Published: 17 June 2026. Checked: 23 October 2026. Coverage: UK. Kind: forecast. Licence: © CITB; figures cited with attribution.
Supports: What a workforce forecast measures. Conflict: CITB is the construction industry's training board and argues for more recruitment and training.
10.Become an apprentice · Department for Education (GOV.UK guide)
Published: 26 September 2012. Last updated: 14 October 2024. Checked: 23 October 2026. Coverage: England. Kind: funding guidance. Licence: Open Government Licence v3.0.
Supports: How apprenticeships work in England. Conflict: None declared in this source record.
11.Free courses for jobs · Department for Work and Pensions
Published: 1 April 2021. Last updated: 29 July 2025. Checked: 23 October 2026. Coverage: England. Kind: funding guidance. Licence: Open Government Licence v3.0.
Supports: Free Courses for Jobs. Conflict: None declared in this source record.
12.Advanced Learner Loan · Student Loans Company (GOV.UK guide)
Published: 16 May 2016. Last updated: 14 October 2024. Checked: 23 October 2026. Coverage: England. Kind: funding guidance. Licence: Open Government Licence v3.0.
Supports: The Advanced Learner Loan. Conflict: None declared in this source record.
13.Open Government Licence v3.0 · The National Archives
Published: undated. Checked: 22 October 2026. Coverage: UK. Kind: licence. Licence: Open Government Licence v3.0.
Supports: Attribution terms for public sector research. Conflict: None declared in this source record.
Contains public sector information licensed under the Open Government Licence v3.0. No publisher endorses TradeTraining or this guide.
Changelog
- : Added practical questions for interpreting the evidence and structured data for the visible FAQs.
- : Added preprint, O*NET, official statistics in development, workforce forecast and three route terms, and pointed several terms at the source we now rely on for them.
- : First publication of the research glossary.