- Artificial intelligence (AI)
- A broad label for computer systems designed to perform tasks associated with intelligence. The label does not, by itself, tell you how a system works or how reliable it is.
- Generative AI
- AI that produces content such as text, images, audio or code in response to input. Generating an answer does not establish that the answer is accurate.
- Large language model (LLM)
- A model trained on large amounts of language-related data to model patterns and generate or process text. Its capabilities and limitations depend on its training and deployment.
- Hallucination / confabulation
- A plausible but incorrect or unsupported output. A confident tone and a realistic-looking citation are not substitutes for checking a claim.
- AI agent
- A system configured to pursue a goal through steps, often using tools. The term covers different levels of autonomy, permissions and oversight.
- Artificial general intelligence (AGI)
- A contested term for broadly capable AI. Definitions and proposed tests differ, so a claim about AGI needs an explicit definition.
- Benchmark
- A defined evaluation task or collection of tasks. A high score describes performance on that evaluation, rather than guaranteeing reliability in every setting.
- Automation vs augmentation
- Automation transfers execution of a task to a system; augmentation supports a human doing it. One job can contain both kinds of tasks.
- Occupational exposure
- An estimate of how tasks in an occupation overlap with AI capabilities. It is not a direct forecast of layoffs.
- Deepfake
- Synthetic or manipulated media that can make a person appear to say or do something they did not. Verification requires context and provenance as well as visual inspection.
These short explanations are AI Affect’s synthesis. For evidence and deeper context, explore our source-linked explainers.