Glossary

Search results

Agentic AI(also: AI agents, autonomous AI agents)

Agentic AI refers to artificial intelligence systems that can plan, make decisions, and execute multi-step actions autonomously to achieve high-level goals, typically by interacting with software, tools, or environments on behalf of a user. Unlike single-turn AI assistants that…

Computer Use Agent(also: CUA, computer-using agent, desktop agent)

A computer use agent is an AI system powered by multimodal large language models that operates a computer by taking screenshots and performing mouse, keyboard, and scroll actions — mirroring the interactions of a sighted user to complete natural language tasks such as booking a…

Context Leakage(also: chat session context bleed)

A failure mode in conversational AI systems in which information from earlier, unrelated chat sessions or turns influences the current response, producing output that blends contexts — for example, mixing VoiceOver (macOS) instructions into a JAWS (Windows) troubleshooting…

Conversational Programming(also: natural language programming, LLM-driven scene modification)

Conversational programming is a paradigm in which users modify software behaviour or digital environments through natural language dialogue with an AI system, rather than through traditional developer-defined controls, menus, or code. In accessibility contexts, conversational…

Conversational Search(also: conversational information retrieval, chat-based search)

Conversational search is an approach to information retrieval in which users interact with a system through natural language dialogue rather than keyword queries, enabling multi-turn exchanges that iteratively refine information needs. For accessibility, conversational search…

Large Language Models(also: LLMs, foundation models)

Large language models are AI systems trained on vast corpora of text data using transformer-based neural network architectures, enabling them to generate, summarize, translate, and reason about natural language. In accessibility contexts, LLMs power conversational assistants…

Mobile Automation(also: mobile workflow automation, trigger-action automation)

Mobile automation refers to the use of scripting tools and platform features on smartphones to automatically perform sequences of actions — such as launching apps, capturing screenshots, invoking APIs, or switching contexts — in response to user-defined triggers. On iOS, Apple…

Prompt Contradiction

A type of large language model failure in which the system disregards explicit instructions or constraints given in the user prompt, producing output that contradicts what was asked. For example, an AI responding with visual instructions like "click the green button" after the…

Runtime Accessibility(also: dynamic accessibility, on-demand accessibility adaptation)

Runtime accessibility refers to the ability to modify, adapt, or enhance the accessibility of a software application or digital environment while it is actively running, rather than through static design-time configurations or developer-authored presets. Traditional…

Semantic Scene Graph(also: SSG, scene graph)

A semantic scene graph is a structured data representation of a 3D or 2D environment that encodes objects, their properties (such as position, size, color, and audio characteristics), and the spatial and hierarchical relationships between them. In accessibility research,…

10 results.