Follow material in an order that builds understanding.
inAiLoading page content...
Unit 05 · assistance that builds agency
Plan for this page
After this unit, you should be able to
Terms you will use
These definitions prepare you for the reading; you do not need to memorize them.Tool use that extends what a person can accomplish while strengthening or preserving the knowledge and judgment required to direct and evaluate it.
Dependence on a tool without sufficient ability to notice failure, continue when it is unavailable, or recover when conditions change.
A planned progression from lighter to stronger support, with opportunities to reduce support and demonstrate capability.
People appropriately rely on many tools: glasses, calculators, search, spelling support, translation, maps, and collaboration. Removing useful assistance can exclude learners or waste effort. The goal is not self-sufficiency from every system, but conscious distribution of capability and responsibility.
Reliance becomes brittle when the learner cannot state the goal, recognize an implausible result, explain a central choice, protect sensitive information, or proceed through a fallback. The relevant threshold depends on consequences: a low-stakes brainstorm and a safety-critical decision require different independent capacity.
Some difficulty is informative: attempting recall reveals what is missing, comparing strategies develops judgment, and revising after feedback connects action to consequence. If AI removes every pause, choice, and error before the learner encounters it, the finished work may conceal limited capability.
Productive effort is not unnecessary hardship. An accessible design can change representation, language, pace, input method, or amount of scaffolding while preserving the intended learning operation. The question is which barrier is irrelevant to the goal and which action is the goal itself.
A capability-preserving workflow asks the person to predict or plan, uses AI for a bounded role, then requires a check, explanation, or application without the original output. Periodic tool-free or tool-light practice can reveal what remains available independently without turning every task into a test.
Recovery completes the design. Keep source materials, intermediate work, export paths, alternative tools, human contacts, and a method for continuing when the system is wrong or absent. Review dependence with evidence from actual performance rather than guilt about using help or enthusiasm about convenience.
Interactive model
The learner chooses goals, asks questions, makes consequential decisions, and can refuse or redirect assistance.
Misconception to correct
Active checks
Explain back and transfer
Source context
Evidence-informed guidance supports retrieval, alternating examples with problem solving and connecting representations; it predates current generative AI and does not quantify long-term dependence on particular AI products.
Supports learner agency, multiple representations and forms of action; accessibility support must not be confused with avoidance of meaningful learning, and individual needs require contextual design.
A broad research programme concerning autonomy, competence and relatedness in motivation; it informs design questions but is not a direct evaluation of generative-AI learning products.
