Deskilling in the Era of Automation: The Hidden Cost of Convenience

Deskilling in the Era of Automation: The Hidden Cost of Convenience

Nikita Pandey and Dr. Himani Grewal

Department of Management, Moradabad Institute of Technology (MIT), Moradabad, Uttar Pradesh, India Corresponding Author E-mail: nikita2003pandey@gmail.com, dr.himanigrewal@gmail.com | Contact: +91-8909453563

Abstract – Artificial Intelligence and automated tools have become significant in the current workplace, where their integration has delivered exceptional efficiency. Yet they raise serious concerns related to employees losing their foundational judgement. This research paper focuses on whether the increasing reliance on automated tools leads to erosion of professional foundations (deskilling) and has diminished their analytical capacity. It then evaluates the actionable strategies and policies that the organisations can formulate in order to reduce these risks. This study adopts a research design that utilises a mixture of both primary data and secondary data. Secondary data is collected from existing works, case studies, journals, and literature reviews, whereas primary data has been gathered using a structured survey through a questionnaire answered by professionals or employees working in various sectors to synthesise a detailed analysis. The study reveals that though the emergence of automated tools has made tasks easier, leading to maximising productivity, it also led to the deskilling of personnel over time. To foster the analytical skills and cognitive efficiency of employees, organisations must ensure the intervention of human resources through upskilling strategies such as the implementation of Human-in-the-loop frameworks.

Index Terms – Artificial Intelligence (AI), deskilling, foundational judgement, hyper-automation, training, workflow automation.

1. Introduction

   Automation has evolved from supporting routine manual labour to shaping cognitive and professional work across multiple do- mains like accounting, engineering, sales, and administration

   [1][3]. This shift has greatly improved efficiency, productivity, speed, and cost reduction, because of which organisations are increasingly relying on AI-enabled tools for decision-making, analysis, reporting, and service delivery [1], [3], [4]. Despite the convenience that improves efficiency in the short term, it can also lead to the gradual erosion of human skills in the long term [1], [5], [6].

   Deskilling is not just the replacement of one task by software. It is a deeper process in which employees lose opportunities to practice work, refine skills, and trust their own expertise [3], [7]. When workers rely on automated outputs excessively, it can make it difficult for them to detect errors, render them less willing to question suggestions, and leave them less confident to act independently if the system fails [1], [8], [9]. Therefore, the concern is not only related to technical proficiency, but also the weakening of the ability to make independent decisions and defend judgments without taking the help of automated tools [9], [10].

   This issue matters because modern organizations increasingly expect humans to remain accountable even when machines are heavily involved in decisions [1], [7]. If workers are responsible for outcomes but have lost the competence to verify or over- ride automation, convenience turns into hidden vulnerability [1], [5], [11]. This paper addresses two objectives: first, examining whether employees who rely on automated tools lose founda- tional judgment over time; and second, analyzing how training

   programs can mitigate that risk by fostering critical thinking, monitoring skills, and independent judgment [3], [4], [11].

2. Literature Review

   The literature on automation and deskilling is broad, establishing that skill decay follows reduced use. Skill decay is generally defined as deterioration in knowledge and ability caused by lack of practice [3], [7]. AI systems cause this by reducing the frequency of skill execution and limiting opportunities for future skill acquisition [3]. Workers may continue to produce acceptable outputs, but the underlying competence required to generate, test, and revise those outputs begins to weaken [1], [7], [12].

   Secondly, deskilling often emerges gradually and remains unnoticed. In accounting case studies, growing reliance on auto- mated software led to complacency, reduced mindfulness, and eventually eroded competence maintenance [1]. This erosion often remains obscure until an operational disruption exposes the dependence [1], [12].

   Furthermore, cognitive offloadingthe delegation of mental work to external systemsexhibits distinct dimensions [4]. Re- cent scholarship distinguishes between assistive offloading (sup- porting cognition while preserving user control), substitutive offloading (replacing mental effort), and disruptive offloading (impairing reflection, self-monitoring, and autonomy) [4], [10]. Over-reliance creates an illusion of competence, causing users to overestimate their understanding while lowering introspection accuracy and critical feedback loops [4].

   Another major stream focuses on automation biasthe ten-

   dency to over-rely on automated recommendations even when contradictory evidence exists [8], [9], [13]. It leads to omis- sion and commission errors across novices and experts alike [13]. At an institutional level, organizational norms often make contesting automated suggestions costly or risky, which reposi-

   Table 1: Survey Variables and Operational Definitions

   Variable Measurement Focus Scale

   tions human disagreement as exceptional behavior and erodes

   Experience

   Total professional work experience Categorical

   confidence [9].

   Sector-specific evidence reinforces these patterns. In aviation, sustained reliance on flight-planning automation degraded pilots manual calculation and estimation capacities [14]. In healthcare,

   Industry Department KnowAI AIuse Double-Check

   Employment sector / industry Specialization / functional area General awareness of AI tools Daily usage of AI / automation Verification of automated results

   Categorical Categorical Binary Binary Likert (15)

   AI-supported diagnostics risk converting clinicians into pas-

   Difficulty-Manual

   Hardship working without tools Likert (15)

   sive machine validators, creating clinical and moral deskilling

   Confidence-Fix

   Confidence troubleshooting errors

   Likert (15)

   risks [7], [11]. In software engineering, code-generation tools

   Team-Solve

   Team problem-solving ability Likert (15)

   raise task completion speedespecially among junior develop-

   Encouraged-Question

   utonomy to override AI advice

   Likert (15)

   erswithout guaranteeing equivalent conceptual mastery [12].

   TrainingType

   Focus: process logic vs. software Categorical

   However, literature notes that deskilling and upskilling can co- occur [3], [12], [15], highlighting the indispensable role of logic- focused continuous training, AI literacy, reflective prompts, and intentional unsupported practice [3], [4], [6], [8], [11], [16] [18].

3. RESEARCH OBJECTIVES

   The study is guided by two specific research objectives:

   1. To understand the long-term effects of using automated tools on the critical judgement and reasoning ability of employees.

   2. To examine the role of training programs in mitigating the risk of deskilling.

      Practice-Manual Frequency of manual skill practice Likert (15)

      1. Primary Data Analysis

         A. Pearson Correlation Analysis

         To assess associations between manual practice, troubleshooting confidence, and task difficulty, Pearson correlation coefficients were computed across respondents (N = 8085), as displayed in Table II.

         Table 2: Pearson Correlation Matrix

         Variable Pair r p-value Significa nce

4. Methodology

Practice-Manual × Difficulty- Manual

-0.242 0.031 Significa

nt (p <

This study adopts a mixed-method research design combining primary empirical survey data with secondary literature synthe- sis.

Practice-Manual × Confidence-Fix 0.390 < 0.001 Significa

nt (p <

Double-Check × Confidence-Fix 0.222 0.048 Significa

nt (p <

Confidence-Fix × Team-Solve 0.608 < 0.001 Significa

nt (p <

1. Research Design and Primary Data

   A structured questionnaire was administered to working profes-

   Encouraged-Question ×

   Confidence-Fix

   Difficulty-Manual × Confidence-

   0.336 0.002 Significa

   nt (p <

   -0.197 0.079 Non-sig.

   sionals across corporate, public, academic, and other sectors. Experience levels ranged from entry-level (< 1 year) to se- nior staff (> 5 years). The questionnaire measured daily AI tool usage, manual difficulty without tools, troubleshooting con- fidence, double-checking behavior, autonomy to question AI recommendations, and enterprise training orientation (logic vs. operational). A total of 85 responses were gathered, yielding 80 85 usable responses across analytical tests after listwise deletion of incomplete entries.

2. Measurement Items

   Table I summarizes the survey variables and corresponding measurement focus.

   Fix trend

   Key correlation findings indicate:

   • Regular manual practice significantly reduces perceived difficulty when tools are unavailable (r = 0.242, p =

     .031) and strongly reinforces troubleshooting confidence (r = 0.390, p < .001).

   • Active verification (Double-Check) correlates positively with troubleshooting confidence (r = 0.222, p = .048), proving human-in-the-loop habits sustain competence.

   • Individual confidence strongly predicts collective team problem-solving efficacy (r = 0.608, p < .001).

   • Autonomy to question AI suggestions elevates individual

confidence (r = 0.336, p = .002).

B. One-Way ANOVA Evaluations

1. Training Type: Comparing training models against manual coping ability showed a strong trend (F = 2.53, p = .063). Employees in organizations focusing deeply on underlying pro- cess logic reported significantly lower difficulty coping without automation (M = 1.25, n = 45) compared to those receiving tool-only training (M = 3.21, n = 14) or no formal training (M = 2.59, n = 34). Confidence-Fix (F = 0.73, p = .536) and Practice-Manual (F = 1.08, p = .364) showed no signifi- cant variance across training groups.

2. Work Experience: A significant effect emerged across experience bands on Difficulty-Manual (F = 3.57, p =

   .018, 2 = 0.121). Tukey HSD revealed that early-career staff (13 years experience) experienced significantly higher diffi-

   task performance tests. Additionally, the cross-sectional nature captures correlational associations rather than direct longitudinal causality.

   VIII. Recommendations

   Table III outlines core organizational strategies for mitigating deskilling risks.

   Table 3: Organizational Strategies to Mitigate Deskilling

   Strategy Operational Focus Basis

   culty performing tasks manually (M = 4.13) compared to se- nior professionals with 5+ years experience (M = 2.49, p =

   .020). This demonstrates a generational cohort vulnerability where newer entrants enter pre-automated environments without establishing foundational manual routines.

3. Industry Sector: Sectoral analysis revealed significant variance in automation dependency (F = 3.93, p = .012, 2 = 0.131). Corporate sector employees reported significantly greater difficulty operating without automated tools (M = 3.31, n = 29) than public sector personnel (M = 2.37, n = 41, p = .012), reflecting deeper algorithmic dependency in pri-

AI Literacy Programs AI-Free Practice Frictional Interfaces Continuous Reskilling Human-in-the-Loop

Educate on algorithmic limits, error modes, and bias Mandatory manual problem- solving sessions

Prompts demanding rationale before tool execution

Logic-centered developmental curricula

Cultural permission to over- ride automated advice

vate enterprises.

1. Discussion and Interpretation

   The empirical and secondary analyses converge on several strate- gic realities:

   • Hands-on Practice as Counter-Deskilling Buffer: Delib- erate manual execution remains the single most impactful safeguard against analytical deterioration (r = 0.39).

   • Cohort Vulnerability: Early-career employees face heightened deskilling risks because their baseline training occurs alongside automated assistants, skipping founda- tional manual struggle.

   • Cognitive vs. Operational Training: Tool-centric oper- ational training accelerates software adoption but leaves employees defenseless during system anomalies. Train- ing emphasizing process logic shields workers from skill degradation.

   • Organizational Autonomy Culture: Empowering em- ployees to challenge algorithmic outputs actively preserves critical reasoning and team capability (r = 0.61).

2. Methodological Limitations

The total sample size (N = 85) is modest, resulting in small cell sizes for certain subgroups (e.g., deep-logic training group n = 45; academic sector n = 5). Outcome measures relied on self-reported Likert items rather than direct psychometric or

IX. Conclusion

Automation delivers unmatched speed and operational efficiency, but excessive cognitive offloading without deliberate skill preser- vation leads to insidious deskilling. When professionals lose the foundational capability to verify, critique, and troubleshoot systems independently, convenience transforms into operational fragility. Protecting foundational judgment requires an organiza- tional shift toward cognitive sustainabilitypairing automation with logic-based training, intentional manual practice, and an institutional culture that empowers human discernment.

References

1. T. Rinta-Kahila, E. Penttinen, A. Salovaara, W. Soliman, and J. Ruissalo, The vicious circles of skill erosion: A case study of cognitive automation, J. Assoc. Inf. Syst., vol. 24, no. 2, 2023.

2. J. Rafner et al., Deskilling, upskilling, and reskilling: A case for hybrid intelligence, Morals & Machines, vol. 1, no. 2, pp. 2436, 2021.

3. S. Morandini et al., The impact of artificial intelligence on workers skills: Upskilling and reskilling in organisations, Informing Sci., vol. 26, pp. 3968, 2023.

4. B. Jose et al., Outsourcing cognition: Thepsychological costs of AI-era convenience, Front. Psychol., vol. 16, Art. no. 1645237, 2025.

5. V. Azarov and V. V. Gorokhov, Cognitive offloading or atrophy? Risks of skill degradation and new forms of dependence, in Proc. IEEE QM&TIS&IT, 2025, pp. 166 168.

6. A. Apida et al., Conceptual review of artificial intelligence over reliance: Opportunities and challenges for employee competencies, Studi Ilmu Manajemen dan Organisasi, vol. 6, no. 4, 2026.

7. C. Natali, L. Marconi, L. D. D. Duran, and F. Cabitza, AI- induced deskilling in medicine: A mixed-method review, Artif. Intell. Rev., vol. 58, 2025.

8. G. Romeo and D. Conti, Exploring automation bias in humanAI collaboration: A review and implications for explainable AI, AI & Soc., vol. 41, pp. 259278, 2025.

9. P. Jovchevski, S. Buijsman, and M. A. Neerincx, What is wrong with automation bias?, Philos. Technol., vol. 39, 2026.

10. D. A. Garcia-Barrios, Epistemic partner or cognitive crutch: A conceptual model of cognitive offloading in humanAI collaboration, IEEE Internet Comput., vol. 29,

    pp. 716, 2025.

11. S. E. A. Tarhouny and A. Farghaly, Deskilling dilemma: Brain over automation, Front. Med., vol. 13, Art. no. 1765692, 2026.

12. K. Crowston and F. Bolici, Deskilling and upskilling with

    AI systems, Inf. Res., vol. 30, pp. 10091023, 2025.

13. R. Parasuraman and D. Manzey, Complacency and bias in human use of automation: An attentional integration, Hum. Factors, vol. 52, no. 3, pp. 381410, 2010.

14. K. Volz and M. Dorneich, Evaluation of cognitive skill degradation in flight planning, J. Cogn. Eng. Decis. Mak., vol. 14, no. 4, pp. 263287, 2020.

15. S. McGuinness, K. Pouliakas, and P. Redmond, Skills- displacing technological change and its impact on jobs: Challenging technological alarmism?, Econ. Innov. New Technol., vol. 32, pp. 370392, 2021.

16. L. Li, Reskilling and upskilling the future-ready work- force for Industry 4.0 and beyond, Inf. Syst. Front., pp. 116, 2022.

17. K. Iyer, Upskilling and reskilling for workforce of the future, 2020.

18. S. K. Bharati and Y. M. Gosavi, The future of work: Up- skilling strategies in the age of automation, Scriptora Int.

J. Res. Innov., vol. 1, no. 5, 2025.