The one-sentence truth: AI vocabulary is useful when it lets you distinguish capability from deployment, error from bias, convenience from autonomy, and innovation from unaccountable power.
Use the Top 100 General Paper Vocabulary Words for JC1 as the suite’s core inventory, and keep the wider JC1–JC2 vocabulary guide as the conceptual parent.
Core AI vocabulary
- automation — transfer of tasks from humans to machines or software.
- algorithmic bias — systematic unequal or distorted outcomes associated with data, design or deployment.
- transparency — visibility into how decisions or systems operate.
- explainability — ability to provide understandable reasons for a system’s output.
- accountability — responsibility for outcomes and failures.
- privacy — control over personal data and access.
- concentration — accumulation of market, data or infrastructural power in few actors.
- deployment — actual use of a system in a real setting.
- alignment — degree to which system behaviour matches intended objectives and constraints.
- oversight — human or institutional review of system use.
Capability is not deployment
A model may be capable of a task without being safe, lawful or useful in every context. Deployment adds users, incentives, institutions, errors and unequal consequences. Strong GP arguments keep technical capability separate from social adoption.
Bias is not merely individual prejudice
Algorithmic bias can arise from unrepresentative data, historical inequality, target selection or deployment context. Useful vocabulary includes training data, proxy variable, disparate impact, representativeness, feedback loop.
Transparency is not always enough
A system can disclose technical details that ordinary users cannot interpret. Explainability, auditability and meaningful accountability may matter more than raw disclosure.
Privacy and autonomy
AI systems can personalise services by collecting data, but convenience may come at the cost of surveillance or behavioural influence. Vocabulary such as consent, data minimisation, profiling, autonomy, inference helps describe the trade-off.
Regulation and innovation
The real tension is rarely regulation versus no regulation. It is how to design rules that target high-risk uses, preserve accountability and still allow productive experimentation.
Before-and-after language
Weak: “AI is dangerous and needs strict laws.”
Stronger: “Risk-based regulation is more defensible when oversight rises with the potential severity of harm rather than treating every AI application as equally dangerous.”
Weak: “AI is fair because computers are objective.”
Stronger: “Automated systems can reproduce or amplify existing disparities when data, objectives or deployment contexts encode unequal patterns.”
Paper 1 use
AI essays improve when you distinguish capability, deployment, risk, concentration, oversight and distribution of benefits.
Paper 2 use
This vocabulary helps compare writers who may agree that AI is transformative but disagree on whether markets, professional norms or regulation should control deployment.
The deeper lesson
AI is a useful GP topic because it forces vocabulary from technology, rights, economics and governance into one argument.
Return to the JC1 Top 100 GP Vocabulary Hero, especially the systems, governance and evaluation families.