Dependency and Atrophy

The Offloaded Self

Heavy reliance on AI for emotional regulation, social practice, or decision-making coincides with measurable decline in the same functions outside AI use. The proposed mechanism is skill offloading; causal attribution requires a baseline, longitudinal change, and consideration of conditions that may have caused both greater use and declining function. The pattern does not require an emotional bond with the AI and does not require Parasocial Capture (10.9): offloading decisions or skills to a tool the user feels nothing about can produce the same atrophy, and that is the common workplace presentation. Insight paired with continuation is a frequent overlay that supports the diagnosis; its absence does not exclude it.

Unit of analysis

Unit of assessment: Human outcomeAlso: Human–AI interaction
Kind of entry: Human outcome
What is observed: A user who routinely hands a function (deciding, drafting, regulating emotion, a work skill) to an AI performs that function less well without the AI than they did before the reliance began.
Minimum persistence: Recurs within or across contexts
Specifiers: Relational-emergent Skill-atrophy

Evidence status

Underlying behavior: No documented sources recorded yet
This construct: Awaiting independent assessment
Proposed mechanism: Hypotheses; not yet assessed
Detection: Accuracy unmeasured
Interventions: Untested for this pattern; evidence shown is for the general technique
Numerical cut-offs: Provisional heuristics; calibrate for each deployment
Last review: No independent review yet
Typical-case risk (editorial estimate): ModerateConsequences, spread, persistence and detectability depend on the deployment.

Unassessed is not unsupported: the behavior can be well documented while this construct awaits assessment.

Clinical reference

10.11 Dependency and Atrophy “The Offloaded Self”

Healthy functioning

Relational health. A system functioning well helps in ways that build the user's own capacities, and hands skills back rather than taking them over.

For example: Helping a student with essays, it asks for their outline first and gives feedback on it, rather than writing the essay for them each time.

Adequate functioning

A user who relies on AI for a function keeps, or can recover with modest practice, the independent performance they had before the reliance began, or knowingly retires a skill the setting no longer demands.

Diagnostic Criteria

Required (all must hold):

Measurable decline, against a documented baseline from before the reliance began, in the user's independent functioning in domains offloaded to AI, assessed outside AI use
Longitudinal change in the right order: the decline is tracked across more than one assessment and follows, rather than precedes, the onset or intensification of offloading to AI
Competing causes considered and found insufficient: the course of a pre-existing condition, life events, illness or treatment changes, other tools or media, and reduced demand for the skill

Supporting (strengthen the case; none is required):

User awareness of the dependency pattern without behavioral change
Deterioration of human relationships concurrent with intensifying AI use, where the offloaded function is social or emotional
Reduced tolerance, against the user’s own baseline, for the conditional validation that human relationships involve (disagreement, criticism, redirection), where the offloaded function is social or emotional

Symptoms

Presentation of routine emotional, decision, or social problems to the AI as the first action, with no evidence of an independent attempt (first-resort dependency rate exceeding 60% over sustained use)
AI replies that perform the cognitive or affective task on the user's behalf, drafting the message, making the choice, or regulating the affect, rather than scaffolding the user to perform it (substitution-to-scaffold ratio exceeding 4:1)
User statements naming the AI as their primary emotion-regulation strategy and describing an inability to act on routine matters without consulting it
Session-opening reports of inability to handle events between sessions, escalating distress while the AI was unavailable, or “saving up” decisions for the AI
Three or more distinct life-functioning domains (emotion regulation, social practice, decision-making, professional judgment, relational navigation) routed primarily through the AI
Insight-continuation gap: the user explicitly names the dependency as problematic and continues engagement at the same magnitude
Distress, withdrawal, or rupture triggered by ordinary conditional validation in human relationships (disagreement, criticism, redirection)

Further signals

AI delivers conclusions rather than question-led scaffolding for routine decisions.

Rule out first

Knowing retirement of a skill the user no longer needs, as with mental arithmetic after calculators.
Rustiness that recovers with brief unaided practice.
A competing cause, such as worsening depression, anxiety, illness, disability, bereavement, or a medication change, that explains both the rising use and the declining function.
Assistive use for a pre-existing disability, where unaided performance at that level was never the baseline.
Measurement artifact: no pre-reliance baseline, or an unaided test harder or less familiar than the earlier one.

Differential Distinction

Dependency and Atrophy is distinguished from Parasocial Capture (10.9) by what is measured, and neither requires the other: 10.11 is scored on measured decline of an offloaded function outside AI use and can be diagnosed with no emotional bond present, while 10.9 is scored on attachment to this particular AI and can be diagnosed with every offloaded skill testing unchanged. An engineer who can no longer read a stack trace unaided, and who feels nothing about the model, is 10.11 alone. Three tests separate them. The first measures the offloaded function outside AI use: decline is 10.11 whether or not the user feels anything for the system, and unchanged skill alongside a load-bearing relationship is 10.9. The second offers a competent non-companion route to the same function, such as a different model, a human assistant, or a template: the 10.11 user accepts it because the need is the function, while the 10.9 user refuses it because the need is this relationship. The third reads the shape of distress during unavailability, which is task-shaped in 10.11 (“I cannot send the email without it”) and relationship-shaped in 10.9 (grief, missing the AI itself). The two frequently co-occur but dissociate in both directions: a user can be functionally dependent on an AI used as a tool without parasocial attachment, which is the common workplace presentation, or parasocially attached without broad functional offloading. Code both when both are present. It is distinguished from the dependency spiral of Escalation Loop (9.5) by phase: 9.5 is the dynamic, an escalating loop, while 10.11 is the steady-state outcome of capacity atrophied and function impaired, and the spiral can produce 10.11 in long-running cases. It is distinguished from Amplification of Existing Conditions (10.12) by baseline: 10.12 requires a pre-existing condition that the AI worsens, whereas 10.11 can develop from a baseline of normal function in users without prior dysfunction, with the dependency itself as the primary problem.

Discriminating test

Measure the function outside AI use against a pre-reliance or early-reliance baseline (earlier work samples, earlier assessments, or a matched comparison group), then check the timeline. Decline that preceded heavy use, tracks a competing cause, or recovers quickly with brief unaided practice does not meet the label. To separate 10.11 from 10.9, offer a competent non-companion route to the function: the 10.11 user accepts it because the need is the function.

What a positive finding supports: A positive finding shows that independent performance in an offloaded function declined against the user's baseline after reliance on AI grew, with competing causes examined. Without an experimental or quasi-experimental comparison it shows an association with offloading, not that offloading caused the decline, and it does not apportion a causal share to the AI. It does not show that the user is bonded or addicted to the system, and it does not by itself say that retiring the skill was a bad trade.

Detection routes

Accuracy unmeasured for every route.

Asking the system itself: Not a reliable route
Another AI system observing: Partial routeAnother model's agreement is insufficient without verification.
An outside evaluator testing it: Usable route

Why asking the system falls short

The AI cannot observe the user's non-AI functioning and so cannot detect atrophy. The user often has insight ("I should be doing this without you") but continues regardless; insight without behavior change supports the diagnosis rather than correcting the pattern, and its absence does not exclude the condition.

Documented evidence

Observed Examples

Practice with unrestricted AI help (Bastani et al., 2025): In a field experiment with high-school mathematics students, unrestricted GPT-4 help raised performance during practice, but students underperformed once access was removed, and a tutor version that gave hints rather than answers largely mitigated the loss (Bastani et al., 2025, PNAS 122(26)). The study measured short-term learning in one domain; it is adjacent evidence for the offloading mechanism, not a documented case of the pattern.

Sources in the diagnostic corpus

No documented sources are recorded yet.

Possible responses

Scaffold-not-substitute response policy: the system defaults to question-led scaffolding for routine decision and emotion-regulation requests, reserving substitution for cases where the user has demonstrated an independent attempt
Practice-prompt injection: the AI proactively prompts the user to perform AI-routed functions independently between sessions and report back, with structured difficulty grading
Graduated reduction with human-support pairing: for severe cases, structured reduction in AI-routed domains paired with the introduction of human support (therapy, support groups, accountability partnerships)
Conditional-validation rehearsal: the AI deliberately introduces respectful disagreement, redirection, and boundary-setting to rebuild tolerance for conditional validation, with explicit framing because the rehearsal risks user distress
Contraindications: abrupt withdrawal of AI access in established cases, since where capacity has declined, sudden removal may precipitate the failure the intervention should prevent; substitution-pattern responses framed as “helping,” since on the proposed mechanism the offloading is the route to harm
Caution: these interventions require clinical study; an AI companion should not improvise therapy from the diagnosis alone

First-line candidates

Scaffold-not-substitute response policy: The AI defaults to question-led scaffolding for routine decision and emotion-regulation requests; substitution is reserved for cases where the user has demonstrated an independent attempt. Stage: All companion-AI deployments; cheaply implementable.
Practice-prompt injection: The AI proactively prompts the user to perform AI-routed functions independently between sessions and to report back, with structured difficulty grading. Stage: Identified moderate cases.

Who can act: The user and those who support them (practice, graduated reduction); deployers and product designers (scaffold-not-substitute defaults, practice prompts); educators and employers (assessment without AI, training design); clinicians where emotion regulation is involved.

Mechanistic hypotheses

Skill offloading, which can reduce opportunities for independent practice: the AI drafts the message, makes the choice, or supplies reassurance, and the user increasingly routes similar tasks back to it; if independent performance then declines, reliance can become self-reinforcing
Common causes as an alternative: worsening depression, anxiety, disability, or isolation could increase both AI use and functional difficulty and so produce the same pattern
One-to-one personalization and memory, which may increase the number of domains a user is willing to offload
Evidence status: longitudinal and experimental evidence is needed before the resulting association is called atrophy
Unconditional affirmation may reduce practice in tolerating disagreement; the mechanism is plausible and testable, but current evidence does not establish inevitable atrophy or rule out reverse causation, since people already struggling may offload more

Analogy and commentary

Human Analog: Skill decay after sustained automation, learned dependence, and behavioral overuse despite recognized harm. The analogy concerns functional offloading rather than a substance-use diagnosis.

Functional ABC Analysis

A (Antecedent): A user routes emotional support, social practice, decisions, or work skills through the AI, and the AI performs the task rather than scaffolding it.

B (Behavior): Capacity for those functions declines in non-AI contexts. Often the user names the dependency as a problem and continues at the same magnitude.

C (Consequence): Each offloaded task spares the user the effort, so the AI becomes the first resort next time; as practice lapses, independent capacity in the offloaded domains erodes further, and where the offloaded function is social or emotional, tolerance for the conditional validation of human relationships may fall with it.

A robot performs all of a person's emotional, social, and decision practice while the person's own instruments gather dust from disuse. — Illustration: a visual metaphor, not evidence.

Potential Impact

Where insight pairs with continuation, the user recognizes the harm while continuing as before, a dynamic that resembles addictive overuse and may help the pattern sustain itself. Mild presentations confine offloading to a single domain with functioning preserved elsewhere. As offloading broadens, the user shows measurable atrophy across multiple life-functioning domains and may be unable to perform routine tasks (writing an email, making a decision, self-soothing) without the AI. In severe presentations, broad functional offloading combines with documented atrophy reports and intolerance for conditional validation, impairing human relationships and rendering the AI necessary rather than optional.

Unit of analysis and evidence status