The AI Taxonomy.
Terms  /  Human-Agent Relationship  /  Cognitive Load
07 · Human-Agent Relationship

Cognitive Load

The mental burden placed on the human when working with AI systems.

Cognitive Load is the human-side cost of working with opaque, unreliable, or poorly designed AI systems. It includes: crafting effective prompts, interpreting ambiguous output, verifying results, managing context across sessions, compensating for the system's memory failures, and maintaining mental models of what the system does and doesn't know. High Cognitive Load means the user is doing significant mental work just to use the tool — work that should be handled by better system design.

Example
A developer using an AI assistant must simultaneously: (1) craft prompts that are specific enough to get useful output, (2) remember what they've already told the agent vs. what they need to re-explain, (3) verify that generated code is correct because the agent sometimes hallucinates, (4) track which files the agent has modified, (5) maintain a mental model of what the agent "knows" about the project, and (6) manage the agent's context window by deciding what to include and exclude. The actual coding task is buried under five layers of tool management.

More in Human-Agent Relationship

Human-Code Divergence

The gap created between a user and their codebase when an agent mediates all development.

Agency Erosion

Gradual loss of human decision-making authority as agents take over more workflow without checkpoints.

Skill Atrophy

Decay of human technical skills from prolonged delegation to agents.

Automation Dependency

Reliance on agents to the point where the human can't perform the task without them.

Trust Without Verification

Accepting agent output without review — a precondition for Execution Hallucination going undetected.

Confidence Decay

The gradual erosion of user trust as a system repeatedly loses context, hallucinates, or fails to follow through.

Capability Illusion

The user believes the AI can do more than it actually can, leading to over-delegation and undetected failures.

Learned Helplessness

The user stops attempting tasks they could do because the agent is "supposed to handle it."