AI deskilling is a structural problem

AI Deskilling

The process by which AI tools, through replacing or mediating human activity, erode humans' core capacities — their ability to think, create, form relationships, and exercise willpower

Structural Problem (in this context)

Ferdman's core argument that deskilling cannot be addressed by individual willpower alone — the design and structure of the socio-technical environment itself either enables or undermines capacity cultivation

Developmental Perfectionism

A neo-Aristotelian philosophical framework holding that human flourishing consists in the competent development and exercise of core human capacities; the basis of the paper's normative argument against deskilling

Core Human Capacities

The set of capacities whose development is constitutive of human flourishing — epistemic capacities, social capacities, the capacity for creativity, and the capacity to will

Epistemic Capacities

The abilities involved in thinking, forming beliefs, and reasoning — including both theoretical rationality and practical rationality (phronesis/practical wisdom)

Phronesis

Practical wisdom; doing the right thing, in the right way, for the right reason — a key component of the epistemic capacities and especially relevant for life-planning

Social Capacities

The moral capacity and the ability to form meaningful relationships of friendship and love — capacities that depend heavily on embodied, intersubjective interaction

Capacity for Creativity

The ability to make unfamiliar combinations of familiar ideas and transform conceptual spaces, without relying on luck, accidents, or mechanical procedures

Capacity to Will (Volitional Capacity)

The meta-capacity to exert effort, overcome difficulty, and persevere toward intrinsically valuable goals — required for developing any of the other capacities

Capacity Impoverishment

The result of inadequate development of the core human capacities, leading to diminished human flourishing and impoverished lives

Skill vs. Habit

A skill involves agential control — the agent initiates the activity; a habit is triggered by something in the environment. Overreliance on AI risks converting skilled behavior into mere habitual responses to triggers

Agential Control

The ability to initiate and adjust an activity as it unfolds — a defining feature of being skilled, as opposed to merely responding to environmental triggers

Habituation

The long, gradual process of practice through which capacities are developed — requires successive trials, learning from failure, and cannot be shortcut by testimony alone

Intersubjectivity

The social dimension of habituation — becoming skilled requires learning from others (mentors, social practices) who instill both the know-how and the shared valuing of the skill

Shared Valuing

The process by which, through intersubjective habituation, a person internalizes the value of a capacity as their own — essential for motivation to continue developing the skill; at risk when AI mediates too many interactions

Embodiment

The requirement that habituation involve physical, embodied engagement — especially important for social and moral capacities, where the physical presence of another person (their gaze, voice) is constitutive of recognizing them as a moral agent

Affordances

Action possibilities formed by the relationship between an agent and their environment — not mere opportunities but invitations that strongly influence what behaviors agents will actually exhibit

Affordance Shrinkage

When the reduction in action possibilities becomes systematic, enduring, and deeply entrenched, actively discouraging users from finding ways to compensate — normatively problematic from a structural standpoint

Capacity-Hostile Environments

Environments that restrict or narrow the field of affordances for capacity cultivation, offering only shallow opportunities for development and discouraging compensation for that shrinkage

Capacity-Conducive Environments

Environments that encourage the development and exercise of human capacities through a rich field of affordances, including opportunities for embodied, intersubjective habituation and volitional exercise

Artificial Personal Assistants (APAs)

LLM-based tools (e.g. ChatGPT, Claude, Gemini) used for life-planning, organizing daily activities, finding purpose, social validation, and interpersonal guidance — the paper's main case study

Life-Planning as Self-Constituting

The view that life-planning is not merely instrumental but identity-constituting — who you are is partly constituted by your life plans — meaning it should not be outsourced to a tool

Plan-Former vs. Plan-Follower

A key distinction in the paper — forming plans requires agential control and exercises multiple capacities; following a plan generated by an APA is more like a habit triggered by the APA's output, bypassing the volitional and epistemic capacities involved in genuine planning

Ironies of Automation (Bainbridge 1983)

The paradox that automating routine tasks leaves humans less prepared for exceptional circumstances, because they are deprived of the routine practice needed to maintain their judgment and skills

Cognitive Patience

The willingness to linger in difficulty — a disposition supported by serendipitous recommendations and productive friction, and undermined by vending-machine AI systems that suppress challenging or unfamiliar content

Serendipity (in AI design)

The deliberate introduction of surprise, novelty, and unexpected recommendations into AI systems — proposed as a capacity-conducive design feature that exercises the volitional capacity and builds cognitive patience

Vending Machine vs. Toolbox Recommender Systems

Vending machines prioritize efficiency and predictability, pigeonholing users into familiar content; toolboxes support exploration, unexpected discoveries, and productive friction — the latter being more capacity-conducive

Virtuous Superheroes

Mitcham's term, used by Ferdman to describe the unrealistic expectation that individuals can simply will themselves to resist deskilling — the structural approach rejects this framing, arguing that environments must be designed to support capacity cultivation

Societal Obligation

Ferdman's conclusion that protecting humans against capacity deskilling is a moral obligation we have to others — not merely a personal choice — requiring attention to institutional, cultural, and governance contexts, not just individual behavior