Beyond Reskilling: Designing the Conditions for Human Agency in AI‑Mediated Work

As Artificial Intelligence (AI) reshapes workflows, reskilling has become the default organisational response. (WEF, 2025). Yet many organisations still see slower adoption than expected, risk‑avoidant behaviours, “checkbox learning,” and uneven performance under pressure; signals that skills programmes alone may not be sufficient. (OECD, 2023)

Here’s a balanced proposition worth considering:

AI transformation is not only a skills problem. It is a conditions problem. (Immordino‑Yang et al., 2019; Thomas et al., 2019)

When work becomes more uncertain, evaluative, and time‑compressed, people can shift into defensive modes, prioritising safety over exploration, which undermines the learning and adaptability that reskilling is meant to enable. (McEwen, 2017; Pessoa, 2008)

This article introduces a leadership lens relevant to senior management, HR/OD, and L&D teams: human agency, and how to design for it at scale. (Bandura, 2006)

Why “more reskilling” isn’t always the answer

Reskilling remains essential. But reskilling can underperform when organisations overlook the neurocognitive and social conditions that make learning usable at work. (Immordino‑Yang et al., 2019; Thomas et al., 2019)

AI‑mediated work often adds uncertainty and opacity (algorithmic decision systems, automated task assignment, unclear accountability) (OECD, 2023), evaluation pressure (maintaining performance while “unlearning”) (McEwen, 2017; Vaughn et al., 2020), and decision density (more judgments and exceptions with less time). (OECD, 2023)

Under these conditions, employees may experience heightened stress, reduced confidence, and diminished learning capacity, meaning training investments struggle to translate into performance. (OECD, 2023; McEwen, 2017)

Leadership implication: if you want sustainable adaptability, you need to build not only capability, but agency. (Bandura, 2006)

What “human agency” means in business terms

Human agency is the capacity to intentionally influence one’s learning, behaviour, and life trajectory through self‑regulation, choice, and purposeful action. (Bandura, 2006)

In organisational life, agency looks like:

• people questioning and sense‑making rather than silently complying, (Thomas et al., 2019)

• people experimenting rather than avoiding unfamiliar approaches, (Di Domenico & Ryan, 2017)

• people articulating reasoning (why/when to trust or override AI outputs), (Thomas et al., 2019)

• people sustaining motivation because they can see a future narrative for themselves as work evolves. (Ashforth & Schinoff, 2016; Immordino‑Yang et al., 2019)

Agency is not a personality trait. In this framework, it is an outcome of aligned workplace conditions: work design, leadership signals, governance, and learning systems. (Sen, 1999; Thomas et al., 2019)

Educational Neuroscience as a leadership lens (not “brain‑based hype”)

Educational neuroscience frames learning as a combination of cognitive, affective, and motivational processes that are biologically constrained and environmentally shaped. (Immordino‑Yang et al., 2019; Thomas et al., 2019)

Practically, learning is not only information transfer; it depends on employees remaining in a regulated, engaged state rather than a defensive one. (Immordino‑Yang et al., 2019)

Prolonged uncertainty, evaluative pressure, and time compression can activate stress responses that impair executive functions relevant to learning, including inhibition, flexibility, and working memory. (McEwen, 2017)

Under sustained stress, people may choose safety over exploration, leading to shallow learning or disengagement. (McEwen, 2017; Pessoa, 2008)

The “agency stack”: four systems leaders must support

Agency can be organised into four interacting neurocognitive systems. (McEwen, 2017; Immordino‑Yang et al., 2019; Di Domenico & Ryan, 2017; Ashforth & Schinoff, 2016)

1) Threat and stress regulation

When uncertainty is experienced as threat, attention and participation shift; learning becomes brittle. (McEwen, 2017; Pessoa, 2008)

2) Executive control

Executive control supports planning, flexibility, and inhibiting previously effective routines; these capacities are strained during continual adaptation and “unlearning.” (Vaughn et al., 2020)

3) Motivation and reward

Motivation supports persistence and exploration and is responsive to autonomy and competence signals in the environment. (Di Domenico & Ryan, 2017; Deci & Ryan, 2000)

4) Meaning‑making and identity integration

Meaning‑making supports coherent professional narratives during disruption and protects identity continuity, especially for experienced workers whose expertise is central to self‑concept. (Ashforth & Schinoff, 2016; Immordino‑Yang et al., 2019)

Critical point: these systems interact. Elevated stress impairs cognition, sustained overload weakens exploration, and identity disruption amplifies threat responses. (McEwen, 2017; Immordino‑Yang et al., 2019)

What leaders (and HR/OD/L&D) can do: interventions at two levels

The framework proposes two interdependent layers:

A) Micro‑level practices that structure day‑to‑day experience, and

B) Meso‑level redesign of systems so agency‑supportive conditions persist through transition. (Thomas et al., 2019; Immordino‑Yang et al., 2019)

A) Micro‑level: structuring employees’ experience of change

1) Contrastive learning sessions (60 to 90 minutes)

Instead of starting with tool training, teams map how decisions were made before, how AI now shapes information/recommendations, and where human judgement remains central. (Menon, 2010; Vaughn et al., 2020)

Measure success via sustained learning confidence, clearer reasoning articulation, and reduced avoidance behaviours. (Thomas et al., 2019)

2) Autonomy‑supportive learning pathways

Offer bounded, meaningful choices aligned to emerging roles with protected learning time to sustain motivation and exploration. (Di Domenico & Ryan, 2017)

Measure success via voluntary learning uptake beyond mandatory requirements and future‑oriented learning narratives rather than compliance completion. (Thomas et al., 2019)

3) Meaning‑making and emotion‑regulation supports

Embed prompts into retrospectives and transition checkpoints to frame uncertainty as effortful learning (not failure) and connect evolving tasks with durable professional purpose. (Immordino‑Yang et al., 2019; Ashforth & Schinoff, 2016)

B) Meso‑level: designing organisations that sustain agency

1) Role and work redesign (as a neurocognitive control intervention)

Clarify how responsibility, judgement, and accountability shift with AI adoption; run task‑control audits to identify where employees retain discretion over sequencing, prioritisation, validation, or exception handling. (McEwen, 2017; Thomas et al., 2019)

Where automation reduces discretion, add compensatory influence (e.g., decision review authority, structured model feedback loops). (McEwen, 2017)

2) Leadership practices as “neurocognitive safety regulators”

Equip managers to detect overload markers (withdrawal, rigidity, silence), pace change announcements away from evaluation cycles and use autonomy‑supportive language that signals options and experimentation. (Pessoa, 2008; Immordino‑Yang et al., 2019)

3) Learning and career infrastructure redesign

Provide visibility rather than prediction: skills clusters, role adjacencies, and examples of lateral movement. This supports employees to map possible futures. (McEwen, 2017)

Critically, governance should explicitly decouple learning exploration data from appraisal or redundancy decisions to reduce evaluative threat and uncertainty‑driven vigilance. (Thomas et al., 2019; Farah, 2005)

Measuring impact without creating surveillance drift

The evaluation approach is deliberately ethical: use behavioural and experiential indicators as proxies, combine/anonymise data, and exclude it from appraisal to avoid triggering evaluative threat. (Thomas et al., 2019; Farah, 2005)

A practical timeline is 6 to 12 months, staged across phases: threat containment → executive/motivational stability → identity integration. (McEwen, 2017; Immordino‑Yang et al., 2019)

Three governance rules worth adopting:

• overall trajectory matters more than individual scores,

• short‑term discomfort is not failure; unresolved stress is,

• workplace improvements should take precedence over individual solutions. (Thomas et al., 2019)

A practical checklist for senior management, HR/OD, and L&D

In your next AI steering meeting, ask:

• Where will uncertainty rise and how will we contain threat? (McEwen, 2017)

• What routines must people unlearn and what scaffolding reduces inhibition load? (Vaughn et al., 2020)

• Where is autonomy shrinking, and what compensatory influence are we designing back in? (Di Domenico & Ryan, 2017)

• Are we protecting learning time or only demanding learning outcomes? (Thomas et al., 2019)

• What identity risks will the experienced staff face and how are we supporting meaning‑making? (Ashforth & Schinoff, 2016)

• What data are we collecting and have we purpose‑limited it away from appraisal in writing? (Farah, 2005; Thomas et al., 2019)

• What does success look like as transfer under pressure, not just training scores? (Thomas et al., 2019)

Closing: an “agency‑first” view of adaptability

Reskilling will remain necessary. But in AI‑mediated work, reskilling becomes effective only when organisations design the conditions that sustain agency: safety, manageable executive demands, autonomy support, and meaning‑making. (Immordino‑Yang et al., 2019; McEwen, 2017)

Don’t replace skills with agency; pair them. Build skills and build conditions that make those skills usable under pressure. (Bandura, 2006; Thomas et al., 2019)

References

Ashforth, B. E., & Schinoff, B. S. (2016). Identity under construction: How individuals come to define themselves in organizations. Annual Review of Organizational Psychology and Organizational Behavior, 3, 111–137. https://doi.org/10.1146/annurev-orgpsych-041015-062322

Bandura, A. (2006). Toward a psychology of human agency. Perspectives on Psychological Science, 1(2), 164–180. https://doi.org/10.1111/j.1745-6916.2006.00011.x

Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227–268. https://doi.org/10.1207/S15327965PLI1104_01

Di Domenico, S. I., & Ryan, R. M. (2017). The emerging neuroscience of intrinsic motivation: A new frontier in self-determination research. Frontiers in Human Neuroscience, 11, Article 145. https://doi.org/10.3389/fnhum.2017.00145

Farah, M. J. (2005). Neuroethics: The practical and the philosophical. Trends in Cognitive Sciences, 9(1), 34–40. https://doi.org/10.1016/j.tics.2004.12.001

Immordino-Yang, M. H., Darling-Hammond, L., & Krone, C. R. (2019). Nurturing nature: How brain development is inherently social and emotional, and what this means for education. Educational Psychologist, 54(3), 185–204. https://doi.org/10.1080/00461520.2019.1633924

McEwen, B. S. (2017). Neurobiological and systemic effects of chronic stress. Chronic Stress, 1, 2470547017692328. https://doi.org/10.1177/2470547017692328

Menon, V., & Uddin, L. Q. (2010). Saliency, switching, attention and control: A network model of insula function. Brain Structure and Function, 214(5–6), 655–667. https://doi.org/10.1007/s00429-010-0262-0

Organisation for Economic Co-operation and Development. (2023). OECD Employment Outlook 2023: Artificial intelligence and the labour market. OECD Publishing. https://doi.org/10.1787/08785bba-en

Pessoa, L. (2008). On the relationship between emotion and cognition. Nature Reviews Neuroscience, 9(2), 148–158. https://doi.org/10.1038/nrn2317

Sen, A. (1999). Development as freedom. Oxford University Press.

Thomas, M. S. C., Ansari, D., & Knowland, V. C. P. (2019). Annual research review: Educational neuroscience: Progress and prospects. Journal of Child Psychology and Psychiatry, 60(4), 477–492. https://doi.org/10.1111/jcpp.12973

Vaughn, A. R., Brown, R. D., & Johnson, M. L. (2020). Understanding conceptual change and science learning through educational neuroscience. Mind, Brain, and Education, 14(2), 82–93. https://doi.org/10.1111/mbe.12237

World Economic Forum. (2025). The Future of Jobs Report 2025. https://reports.weforum.org/docs/WEF_Future_of_Jobs_Report_2025.pdf