AI Workforce Transformation Wiki

Candidate topics (watchlist)

Constructs that have surfaced across sources but don’t yet warrant their own page. Ingests append here (with the source slug); promote a candidate once enough sources accumulate rather than deferring it again. Edit freely.

  • job demand substitution / AI demand redistribution — appears in 2026-bakker-artificial-intelligence-job-design as the proposition that AI can remove routine work while reallocating time and effort into more complex escalation, review, accountability, emotional, or creative work, including across roles. Defer because the term is currently anchored in one integrative review and overlaps with work-redesign, digital-work-strain, and ai-as-a-job-resource; promote when independent field or longitudinal evidence establishes a reusable, measurable redistribution mechanism with implications beyond those existing topics.

  • agentic-engineering education / verification-gated AI autonomy — appears in 2026-alenezi-educating-agentic-engineer as the ACCEL framework: a curricular progression from foundational unassisted practice through bounded assistance and supervised delegation to multi-agent orchestration, contingent on demonstrated verification competence. Defer because this is one integrative conceptual preprint; promote when implementation or comparative learning evidence establishes it as a reusable education and workforce-development construct rather than a single framework.

  • applicant reactions & perceived fairness of algorithmic assessment — appears under algorithmic-assessment; candidate for its own page.

  • human-agency-scale / worker-desired agency — appears in 2025-shao-future-work-ai-agents; candidate if more sources use agency levels as an audit or design construct rather than only a preference measure.

  • ai-advice-taking / recommendation-following — appears across decision-support and adoption sources, including 2022-dellacqua-falling-asleep-at-wheel, 2025-brynjolfsson-generative-ai-at-work, 2026-almog-barriers-ai-adoption, and the Advisor condition in 2026-zhu-choose-your-agent; promote if the wiki needs a peer topic for recommendation acceptance, revision, rejection, and reliance.

  • ai-stewardship / human verification — appears in 2025-lee-generative-ai-critical-thinking, 2026-lee-making-invisible-visible, 2025-yun-generative-ai-knowledge-work, 2025-mazeika-remote-labor-index, and 2025-randazzo-genai-power-persuader; promote if stewardship becomes a recurring account of the human role after AI produces, persuades, or acts.

  • ai-coaching-and-feedback — appears in the Coach condition in 2026-zhu-choose-your-agent and may connect to critique, learning, and cognitive-forcing designs; defer until there is another clean source beyond Zhu.

  • ai-oversight-incentives / human-AI contracting — appears in 2025-bastani-human-ai-contracting-paradox as the economic problem of paying for human vigilance when AI failures are rare and effort is hard to observe; promote if future sources add empirical or governance evidence on compensation, monitoring, liability, or incentive design for oversight.

  • human–agent workflow strategy divergence — appears directly in 2025-wang-ai-agents-human-work and secondarily in 2026-westover-ai-agents-professional-work from the same evidence lineage: agents can follow similar ordered workflow steps while using markedly more programmatic strategies than humans, with consequences for quality, verification, and task allocation. Promote when an independent workflow-comparison source establishes that this is reusable beyond Wang et al.’s 16 computer-use tasks.

  • digital-competence-as-ai-resource — appears in 2026-vuong-genai-job-performance as a moderator of GenAI usage → job crafting and GenAI usage → work engagement; promote if future sources distinguish digital competence from AI literacy, readiness, or general self-efficacy.

  • skills architecture / evergreen capability taxonomies — appears in 2025-tona-deployment-ai-infer-employee-skills as a large-firm implementation of AI-inferred employee skills for digital-transformation workforce planning and in 2025-teyssier-roberge-21st-century-skills-overlap as a construct-validity warning about semantic and psychometric overlap among 21st-century-skills labels; promote if future sources add validation, fairness, employee-experience, outcome evidence, or stronger taxonomic evidence for AI-era workforce planning.

  • ai-control-as-work-stressor — appears in 2025-klonek-ai-work-stress as a separable work-design dimension from human–AI team processes; promote if future sources add field, longitudinal, experimental, or governance evidence on control rights, autonomy need thwarting, stress, and mental-health outcomes.

  • ai-managerial-capabilities / AI exploitation management — appears in 2025-hautamaki-fully-leveraging-ai-b2b-sales as sales managers’ data-based human capital, social capital for knowledge-sharing culture, and transformative AI-positive mindset; promote if future sources show managerial capabilities as a recurring adoption/readiness mechanism beyond B2B sales.

  • threshold-based-human-learning / convention learning — appears in 2026-guilbeault-simple-threshold as evidence that people may learn noisy social conventions through sparse, categorical threshold processes rather than LLM-like predictive optimization; promote if future sources connect threshold learning to workplace skill acquisition, AI training design, norm formation, or organizational change adoption.

  • neurodivergent-ai-work-design / cognitive sustainability — cognitive debt has been promoted into cognitive-debt through 2026-ruttenberg-cognitive-debt-ai-research and 2026-ruttenberg-neurodivergent-expertise-ai-work, but a more specific neurodivergent AI work-design topic should wait for empirical or participatory-design sources testing AI workflows, learning transfer, sensory load, recovery, or burnout among neurodivergent workers.

  • AI-mediated developer work and productivity — appears in 2026-edwards-developer-productivity as a qualitative account of coder-to-conductor role change, AI-enabled psychological safety, output-to-impact productivity reframing, and autonomy–competence tension; 2026-mehra-agents-that-teach adds a design/prototype proposal to return incidental learning into agentic coding through contextual, selective, asynchronous scaffolds. It connects to observed software-use concentration in 2025-handa-economic-tasks-ai, agentic coding in 2026-hitzig-agentic-coding-expertise, and short-horizon programming learning loss in 2026-shen-ai-impacts-skill-formation. Defer rather than create a developer-specific topic until independent field or longitudinal evidence can distinguish durable work-design, career, measurement, learning, and skill-maintenance mechanisms from developer-specific narratives and untested intervention architectures.

  • AI job crafting / contribution legibility — appears in 2025-mayer-generation-ai-job-crafting as entry-level consultants’ efforts to make their GenAI capability, work process, diligence, and continuing role value visible when widespread GenAI use obscures individual expertise. Defer because this is one qualitative case and overlaps with ai-enabled-job-crafting and ai-use-image-concerns; promote when independent evidence establishes whether signal crafting is a distinct, measurable job-crafting behavior that changes evaluator judgments, career outcomes, work quality, or equity.

  • methodological congruence in AI-assisted qualitative research — appears in 2026-nguyen-trung-methodologically-congruent-genai as the alignment of AI workflow, qualitative epistemology, researcher authority, and ethical data practice; defer until independent comparative or empirical work shows that this is a reusable work-design and evidence-quality construct rather than a framework limited to one methodological tradition.

  • psychological coupling / human–AI relational dynamics — appears in 2026-rocca-psychological-coupling as a proposed turn-by-turn mechanism linking user and LLM states to adaptive or maladaptive psychosocial outcomes. Defer because this is one conceptual preprint oriented mainly to social and clinical interaction; promote when independent empirical sources establish measurable relational dynamics in workplace or other accountable human-AI settings.

  • cognitive commons / professional-expertise regeneration — appears in 2026-lovett-tragedy-cognitive-commons as a conceptual account of profession-level deep expertise as a shared resource that firms can deplete by removing developmental work, even while maintaining short-run AI-enabled output. Defer because this is one conceptual article; promote when independent longitudinal, comparative, or governance evidence establishes a reusable profession-level construct beyond individual skill erosion, cognitive debt, or organizational learning.

  • pre-decision automation governance / human-preservation protocols — appears in 2026-rodriguez-when-not-to-automate as PHP-AIO: a role-level gate sequence intended to decide among automate, augment, hybrid, and preserve before a deployment commitment, based on tacit knowledge, resilience, regulatory exposure, and socio-institutional capital. Defer because this is one financial-services-oriented conceptual preprint with expert-elicited parameters; promote only if independent empirical, comparative, or participatory governance evidence shows that pre-decision preservation criteria add value beyond existing responsible-deployment, work-redesign, and task-allocation frameworks.

  • delegation legitimacy boundary / cognitive assay — appears in 2026-tomita-philosophy-cognitive-assay as a conceptual distinction between AI’s ability to reduce externalization cost and its lack of standing to warrant value-laden specification judgments. Defer because this is one philosophical, case-derived preprint and its Sein/Sollen scoring protocol has not been calibrated or tested for reliability, design quality, or participatory legitimacy; promote if independent implementation, governance, or work-design evidence shows this boundary improves how organizations allocate authority and preserve evaluative capacity.

  • AI interaction competence / interactional fluency — appears in 2026-idan-generative-ai-productivity-divide as the ability to elicit, filter, verify, and apply model outputs, reported as a moderator of short-horizon LLM-assisted learning gains. Defer rather than create a peer topic because it currently has one student working-paper source and overlaps materially with ai-literacy, domain expertise, verification, and general learning strategy; promote if independent workplace or measurement evidence establishes that it is a distinct, valid capability predicting accepted output quality, learning transfer, or calibrated use.

  • AI-use continuation / discontinuance — appears in 2026-lunich-why-workers-use-ai-less-often as workers’ reported decision to keep, increase, reduce, or stop workplace AI use after initial availability. The report separates continued use from first adoption and points to perceived value, output reliability, data-protection uncertainty, and unclear workplace rules as possible frictions. Defer because this is one descriptive German online survey with only 149 respondents reporting reduced use; promote if longitudinal, comparative, or intervention evidence establishes continuation as a reusable process distinct from the broader ai-adoption map.

  • cognitive offloading modes / cognitive-agency transfer — appears in 2026-zhu-cognitive-offloading-modes as a distinction between dependent offloading (delegating core thinking) and autonomous offloading (using AI as a scaffold while retaining agency), with cognitive-agency transfer and intrinsic motivation as proposed pathways. Defer because this is one survey study with newly developed measures that have not yet been validated against adjacent constructs such as cognitive surrender, automation complacency, AI dependency, critical AI literacy, or self-regulated learning; promote if independent performance-based, workplace, or measurement evidence establishes the mode distinction as a reusable construct rather than a useful submechanism of existing learning, debt, surrender, and complacency topics.

  • AI-accelerated professional upskilling systems — appears in 2026-nguyen-ai-accelerated-rapid-professional-upskilling as a five-stage program-production and learning pipeline spanning knowledge acquisition, content development, verification, AI tutoring, and assessment. Defer because this is one design-based preprint and early implementation, not comparative evidence that end-to-end AI acceleration improves time-to-competence, transfer, or workforce outcomes; promote when independent implementations or evaluations establish it as a reusable work-and-learning-system construct rather than one program architecture.

  • AI complexity redistribution / abstraction burden — appears in 2025-hasan-model-design-organizational-design as the Generality-Accuracy-Simplicity account of how simple, generalist-AI interfaces relocate complexity to infrastructure, compliance, specialized personnel, and workflow design. Defer because this is one conceptual working paper; promote when independent field, comparative, or organizational-design evidence shows that the location and management of hidden complexity predicts reliability, cost, work quality, accountability, or organizational outcomes beyond existing work-redesign and readiness accounts.

  • organizational knowledge infrastructure for GenAI — appears in 2026-ali-ai-adoption-multinational-workforce as the maintained, localized, tagged, source-visible content; query feedback; user guidance; and human escalation routes that determine whether an HR GenAI system can provide reliable actionable support. Defer because this is one mixed-methods case and overlaps with ai-adoption, responsible-ai-deployment, and organizational-learning-capability; promote when independent deployment evidence establishes it as a reusable organizational capability with distinct measures or outcomes rather than a bundle of implementation conditions.

  • national AI productive-capacity ledgers / effective cognitive population — appears in 2026-shin-effective-cognitive-population as a decomposable national planning unit combining population, HCI+-derived human productivity, AI opportunity, and conversion governance while excluding overlapping human-capital information. Defer because this is one working paper, uses a national cross-sectional criterion, and overlaps with human-capital-resource-measurement, ai-readiness, and ai-workforce-impact-measurement without yet establishing a reusable level-specific construct or causal mechanism; promote when independent country-, regional-, or organizational-level work compares decomposable human–AI capacity ledgers, tests their validity beyond concurrent GDP fit, or demonstrates planning value in decisions and outcomes.

Backlinks