Outcomes: what the engagement strengthens or weakens

Supports Learning

• Select

  Retention is not the same as fluency

  What it established. Retrieving material produces better long-term retention than restudying it, even though restudying feels more productive and learners predict the opposite.

  What it licenses here. Why learning is measured after a delay and without assistance, rather than during the task. It is also the reason immediate performance is a poor proxy: the feeling of learning and the fact of it come apart, and AI is very good at producing the feeling.

  Roediger & Karpicke (2006). Test-Enhanced Learning. Psychological Science, 17(3), 249-255. Source

• Select

  Improvement comes from effortful practice, not exposure

  What it established. Expert performance is built by extended deliberate practice with feedback on tasks at the edge of current ability, rather than by accumulated experience alone.

  What it licenses here. The model's insistence that time spent is not development. It does not license calling any AI workflow deliberate practice; that is an empirical question nobody here has answered.

  Ericsson et al. (1993). The role of deliberate practice in the acquisition of expert performance.. Psychological Review, 100(3), 363-406. Source

• Select

  Not all learning outcomes are the same kind

  What it established. Educational objectives differ in kind, from recall through to evaluation and synthesis, and an assessment aimed at one says little about the others.

  What it licenses here. Why the model refuses a single learning score. A task where AI helps recall and one where it displaces evaluation are not the same result with different signs.

  Bloom (1956). Taxonomy of Educational Objectives: The Classification of Educational Goals. Longmans, Green. Book (Longmans, Green). DOI did not resolve.

Supports Judgment

• Select

  Judgment sits on a continuum, not a dichotomy

  What it established. Cognition ranges between intuitive and analytical poles, and the effective mode depends on how well the task's structure matches the mode used.

  What it licenses here. The learning account underneath the whole model, and the reason no mode is universally correct. Task structure, not virtue, decides.

  Hammond (1996). Human Judgment and Social Policy. Oxford University PressNew York, NY. Source

• Select

  Reflective override is measurable

  What it established. People differ measurably in the tendency to override an intuitive answer that is wrong, and that tendency predicts decision quality independently of ability.

  What it licenses here. One concrete, non-self-report handle on the judgment outcome. Its narrowness is the point: it is a small measure, not a proxy for judgment as a whole.

  Frederick (2005). Cognitive Reflection and Decision Making. Journal of Economic Perspectives, 19(4), 25-42. Source

• Select

  Friction can restore evaluation

  What it established. Cognitive forcing functions — design choices that interrupt the acceptance of an AI suggestion — reduced overreliance, at a cost in speed and in how much people liked the system.

  What it licenses here. The nearest thing in the literature to a test of the cultivability claim, and the source of the prediction that forcing evaluation narrows the gap. It also names the trade: the interventions that worked were the ones people preferred least.

  Buçinca et al. (2021). To Trust or to Think. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW1), 1-21. Source

Supports Agency

• Select

  Ownership comes through control, knowledge and investment

  What it established. Feeling that something is yours arises from controlling it, coming to know it intimately, and investing yourself in it, and that feeling changes how responsibly you treat it.

  What it licenses here. Agency as something that can quietly erode rather than being lost in one decision. All three routes run through the parts of work AI most readily absorbs.

  Pierce et al. (2003). The State of Psychological Ownership: Integrating and Extending a Century of Research. Review of General Psychology, 7(1), 84-107. Source

• Select

  Where a person locates control is stable and consequential

  What it established. Whether people attribute outcomes to their own action or to outside forces is a stable expectancy that shapes how they respond to feedback.

  What it licenses here. Why agency is treated as an outcome that can move rather than a fixed trait, while acknowledging it moves slowly.

  Rotter (1966). Generalized expectancies for internal versus external control of reinforcement.. Psychological Monographs: General and Applied, 80(1), 1-28. Source

Supports Voice

• Select

  A clear self-concept is measurable and consequential

  What it established. The clarity and internal consistency of a person's self-concept is a measurable individual difference, distinct from self-esteem.

  What it licenses here. Voice as more than style. If what you produce stops resembling how you think, the model treats that as a real cost rather than an aesthetic one.

  Campbell et al. (1996). Self-concept clarity: Measurement, personality correlates, and cultural boundaries.. Journal of Personality and Social Psychology, 70(1), 141-156. Source

• Select

  Goals are formed during composition, not before it

  What it established. Writing is recursive rather than transcriptive: goals are discovered and revised while writing, not settled in advance and then rendered.

  What it licenses here. Why delegating composition delegates some goal-setting with it. This identifies where the cost would occur; it does not claim that AI-assisted writing always incurs it.

  Flower & Hayes (1981). A Cognitive Process Theory of Writing. College Composition & Communication, 32(4), 365-387. Source

Supports Flourishing

• Select

  Some motivations sustain and others deplete

  What it established. Behavior supported by autonomy, competence and relatedness sustains itself and predicts wellbeing, while externally controlled motivation does not.

  What it licenses here. The bound on this outcome. Flourishing is included because the model would be dishonest to stop at productivity, and it is marked as the most distal and least measurable thing here.

  Deci & Ryan (2000). The "What" and "Why" of Goal Pursuits: Human Needs and the Self-Determination of Behavior. Psychological Inquiry, 11(4), 227-268. Source

• Select

  Absorption in demanding work is its own reward

  What it established. Sustained engagement in tasks that match challenge to skill is experienced as intrinsically rewarding and is a component of a life people rate as good.

  What it licenses here. Why the model does not treat 'the task got easier' as an unambiguous gain. What productive friction protects is partly this.

  Csikszentmihalyi (1990). Flow: The Psychology of Optimal Experience. Harper & Row. Trade book (Harper & Row). The tested DOI resolves to a 2014 collected-works volume, not Flow itself.