Can safer work become worse work?

Introduction

Automating dangerous work can remove some of the worst features of industrial life. Fewer workers need to enter unstable mines, handle toxic chemicals, climb hazardous structures or perform physically damaging repetitive labour. For supporters of an AI-enabled future of abundance, that matters because reducing injury, illness and premature death is one of the clearest ways technology can expand human flourishing.

But safer work is not automatically better work. A factory, warehouse, transport network or energy facility can become physically safer while simultaneously becoming more monitored, more monotonous and less dependent on human judgement. Workers may move from active problem-solving roles into passive supervisory positions where software makes most decisions and humans intervene only when something goes wrong. In extreme cases, people become attendants to machines rather than skilled participants in production.

This creates an important challenge for the broader AI bloom vision. If advanced AI and robotics eliminate dangerous labour but also hollow out autonomy, expertise and meaning, then part of the promised gain is lost. The question is not only whether technology can make work safer. It is whether it can make work safer without stripping away the skills and agency that help people develop competence, status and purpose.

Why physical safety and job quality can diverge

Workplace safety and job quality often improve together, but they are not the same thing.

A mining worker who no longer enters unstable tunnels is clearly safer. A warehouse employee who no longer spends hours lifting heavy loads may face fewer long-term injuries. A remote operator controlling industrial equipment from a control room avoids many physical hazards that earlier generations accepted as normal.

Yet these gains can arrive alongside a different set of losses. The worker may now spend most of the day watching screens, responding to automated alerts and following software-generated instructions. Instead of exercising judgement throughout a task, they may only be permitted to intervene during rare exceptions.

The OECD has repeatedly highlighted this tension. AI can improve occupational safety and some aspects of job quality, while also creating risks around loss of agency, excessive monitoring and reduced worker control.[OECD]oecd.orgartificial intelligence job quality and inclusiveness a713d0adArtificial intelligence, job quality and inclusiveness11 Jul 2023 — This chapter reviews the current empirical evidence of the effect…[OECD]oecd.orgAI and workAI can bring many benefits to the workplace such as higher productivity, improved job quality and stronger occupational safety…

This distinction matters because people generally value more than wages and safety alone. Research on work quality consistently finds that autonomy, skill development, social recognition and the ability to exercise judgement are major contributors to job satisfaction. A role can become physically easier while becoming psychologically thinner.

In the optimistic AI bloom scenario, dangerous labour is not merely removed. Humans are supposed to be freed for more creative, skilled and meaningful activities. Deskilling represents a warning that technological progress does not automatically move in that direction.

How automation can hollow out expertise

Deskilling occurs when technology reduces the need for workers to develop, maintain or exercise skills that were previously central to their role.

Historically, this has appeared in many industries. Skilled crafts were sometimes broken into simpler standardised tasks. Digital systems later automated portions of clerical, technical and professional work. AI extends this process because it can increasingly perform cognitive tasks as well as physical ones.

Several mechanisms repeatedly appear.

Skill atrophy through reduced practice. Workers may still possess knowledge in theory, but lose competence because software performs most of the relevant actions. Safety specialists have long warned that automation can leave operators unable to respond effectively when systems fail because they rarely perform the underlying task themselves.[aubreydaniels.com]aubreydaniels.comis deskilling a threat to safety in your workplaceMaybe you've heard the term de-skilling. It refers to the loss of knowledge or skills on…Read more…

Conversion of judgement into procedure. Activities once requiring experience become structured workflows with predefined options. Instead of diagnosing problems independently, workers follow software prompts and escalation rules.

Loss of apprenticeship pathways. Many professions rely on juniors learning through repeated exposure to routine work. When AI or automation absorbs those tasks, training pipelines can weaken. Concerns about disappearing learning opportunities have appeared in fields ranging from law and software development to technical operations.[Medium]zephoria.medium.comDeskilling on the JobDeskilling on the Job - danah boydApril 21, 2023 — We may be fine with deskilling junior lawyers now, but how do we generate future…Published: April 21, 2023

Concentration of expertise. Knowledge can migrate from frontline workers into software systems, engineering teams or central management functions. Workers continue operating the system but have less influence over how it functions.

The result is not necessarily unemployment. Many workers remain employed. The concern is that their role becomes narrower, making them easier to replace and less able to develop deeper expertise over time.

The rise of the supervisor who mostly watches

One of the most common outcomes of safety-oriented automation is the creation of monitoring roles.

Instead of directly performing dangerous work, employees supervise robots, inspect dashboards, review camera feeds or monitor sensor networks. This often delivers genuine safety benefits. A worker sitting in a remote operations centre is less likely to be crushed by heavy machinery than a worker standing beside it.

The problem is that monitoring can become psychologically demanding in a different way.

Human attention is poorly suited to long periods of passive vigilance. Decades of research on automation have found that operators often struggle when they are asked to remain alert while systems handle nearly everything automatically. Most of the time there is nothing to do. Then, during a failure, rapid intervention becomes critical.

This creates a paradox. The worker is expected to remain highly skilled, but the job gives few opportunities to exercise those skills.

Industrial control rooms, autonomous vehicle supervision systems and highly automated logistics environments increasingly confront this challenge. When software performs the routine work, humans are left managing exceptions. Yet expertise often develops through routine engagement, not only through emergencies.

The risk is that organisations retain nominal human oversight while gradually eroding the practical competence required for meaningful oversight.

When safer workplaces become surveillance workplaces

Another tension emerges when safety technologies merge with algorithmic management.

Many AI systems introduced for safety purposes also generate detailed data about worker behaviour. Sensors can track movement, location, speed, productivity, fatigue, biometric indicators or compliance with procedures. Employers may argue that such monitoring helps prevent accidents, optimise workflows and identify risks.

Sometimes it does.

However, researchers and labour organisations increasingly warn that the same infrastructure can become a system of continuous surveillance. OECD research on algorithmic management notes both productivity benefits and growing concerns about detrimental effects on workers.[OECD]oecd.orgAI and workAI can bring many benefits to the workplace such as higher productivity, improved job quality and stronger occupational safety…

Studies of workplace monitoring have identified risks including reduced autonomy, increased stress and a perception of constant evaluation.[JRC Publications]publications.jrc.ec.europa.euJRC Publications Electronic Monitoring and Surveillance in the WorkplaceJRC PublicationsElectronic Monitoring and Surveillance in the WorkplaceNovember 22, 2021 — by K BALL · Cited by 200 — When remote working…Published: November 22, 2021

The distinction between assistance and control can become blurred.

A wearable device might warn a worker about dangerous fatigue levels. It might also feed performance data into automated disciplinary systems. A route-optimisation tool might reduce driving risks. It might also pressure workers to follow machine-generated schedules with little discretion.

The concern is not simply privacy. It is the transformation of work into a series of measurable outputs where human judgement becomes secondary to compliance with algorithmic instructions.

Several recent analyses of algorithmic management argue that excessive monitoring can contribute to deskilling by reducing workers’ ability to decide how tasks should be performed.[ifow.org]ifow.orgmaking algorithmic management safe for workers new regulation is neededMaking Algorithmic Management safe for workers28 Jul 2023 — AM's automation of tasks can lead to deskilling, where workers see a fall in…[National Employment Law Project]nelp.orgNew Report Details Employers' Harmful Use of Digital…15 Jul 2025 —… dangers posed by digital surveillance and automated decision sy…

The hidden long-term problem: who still knows how the system works?

Deskilling creates a deeper issue that may not become visible until systems fail.

Highly automated environments often rely on a relatively small group of engineers, designers and technical specialists who understand the underlying systems. Frontline workers increasingly interact with interfaces rather than mechanisms.

For routine operations, this may be efficient.

But rare failures can expose the weakness of overly automated organisations. When unusual conditions emerge, workers may lack the practical understanding needed to diagnose problems independently. Instead of skilled operators supported by technology, organisations can end up with workers who are dependent on technology.

This creates a form of fragility.

In safety-critical sectors such as energy infrastructure, transport, healthcare and industrial production, maintaining human expertise is often valuable even when machines perform most routine tasks. Organisations need people who understand why a system behaves as it does, not merely how to follow its outputs.

The issue becomes even more important in a future where advanced AI systems may manage increasingly complex parts of the economy. If human capabilities steadily erode while machine capabilities expand, society could become more dependent on systems that fewer people truly understand.

Critics of AI optimism often point to this possibility. A civilisation may become technologically powerful while losing distributed human competence. The challenge is not only economic. It concerns resilience and human agency.

Ways to preserve autonomy, mastery and human judgement

The existence of deskilling risks does not mean automation inevitably produces worse work.

Many researchers and policymakers now focus on how technology can augment workers rather than merely constrain them. OECD work on AI and the future of work increasingly emphasises worker empowerment, skills and job quality alongside productivity and safety.[OECD AI]oecd.orgalgorithmic management in the workplace 287c13c4 enAlgorithmic management in the workplaceby A Milanez · 2025 · Cited by 48 — Algorithmic management – the use of software, which may in…[OECD]oecd.aialgorithmic management in the workplace6 Feb 2025 — The survey offers unprecedented insights into algorithmic management in the workplace, its perceived impacts and firm-level…

Several design principles appear repeatedly.

Use automation to assist judgement rather than replace it

The strongest augmentation models leave meaningful decisions with humans.

Instead of generating instructions that workers must follow, AI systems can provide analysis, forecasts, risk warnings and recommendations that support decision-making. Workers remain active participants rather than passive recipients of commands.

This approach tends to preserve learning because people continue exercising judgement.

Protect opportunities to build expertise

Training systems often rely on routine work that appears inefficient from a narrow automation perspective.

Removing every beginner task may improve short-term productivity while weakening long-term skill formation. Organisations need deliberate pathways through which workers can acquire practical experience, understand system behaviour and progress toward more advanced roles.

Without these pathways, industries can struggle to reproduce expertise across generations.

Give workers influence over technological design

Research on automation frequently finds better outcomes when workers participate in deployment decisions rather than having systems imposed upon them. Frontline employees often possess knowledge about workflows, failure modes and operational realities that designers overlook.[ACM Digital Library]dl.acm.orgThis, in turnACM Digital LibraryExploring Labor Relations in Workplace Automation and…11 May 2024 — The developers' efforts to exert more control c…Published: May 2024

Worker involvement can help ensure that automation removes hazards without unnecessarily removing discretion.

Measure success beyond injury rates

A factory that halves workplace injuries has achieved something valuable. But injury reduction alone should not define success.

Organisations can also track:

A safer workplace that leaves workers disengaged, easily replaceable and permanently monitored may solve one problem while creating another.

What this means for the AI bloom vision

The optimistic case for advanced AI is not simply that machines perform more work. It is that humanity gains freedom from drudgery, danger and unnecessary suffering while expanding its capacity for learning, creativity and achievement.

Deskilling sits directly at the centre of that debate.

If automation removes hazardous labour but turns workers into tightly monitored attendants of opaque systems, then technological progress may improve safety while undermining some dimensions of human flourishing. If, instead, AI systems remove the most dangerous and repetitive elements of work while helping people develop higher levels of competence, judgement and creativity, the outcome looks very different.

The distinction matters because abundance is not only about material output. A future with greater wealth, lower injury rates and extraordinary technological capability could still disappoint if most people lose opportunities to exercise skill, develop mastery or shape their own work.

For advocates of an AI-enabled long-term future, the challenge is therefore larger than replacing dangerous jobs. It is designing institutions, workplaces and technologies that preserve meaningful human agency even as machines become dramatically more capable. The success of safer automation may ultimately be judged not only by how many injuries it prevents, but by whether it leaves people with more room to grow rather than less.

Amazon book picks

Further Reading

Books and field guides related to Can safer work become worse work?. Use these as the next step if you want deeper reading beyond the article.

Book

The Glass Cage

By Nicholas Carr

Directly addresses how automation can make work safer or easier while eroding human skill and judgement.

See on Amazon

Book

Human + Machine

By Paul R. Daugherty, H. James Wilson

Focuses on designing roles where humans retain judgement and machines provide support.

See on Amazon

Book

The Shallows

By Nicholas Carr

Supports the theme that convenience can weaken attention, mastery and independent capability.

See on Amazon

Book

Automation and the Future of Work

By Aaron Benanav

Examines how automation can worsen job quality unless institutions shape the transition.

See on Amazon

Browse more on Amazon:The Glass CageHuman + MachineThe Shallows

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

UsingUSA USA USAUKCanadaAustraliaIreland

Example eBay listing

The Wild Robot 2024 Movie Poster Wall Art Print A5 A4 A3 A2 A1 Maxi 2785

SearcheBay.co.uk: robot poster

Browse similar oneBay.co.uk

Example eBay listing

The Wild Robot 2024 Teaser Movie Poster

SearcheBay.co.uk: robot poster

Browse similar oneBay.co.uk

Example eBay listing

Robot Framed Art Print Framed Wall Art Poster Canvas Print Picture

SearcheBay.co.uk: robot poster

Browse similar oneBay.co.uk

Example eBay listing

The Wild Robot 2024 Movie Poster Wall Art Print A5 A4 A3 A2 A1 Maxi 2784

SearcheBay.co.uk: robot poster

Browse similar oneBay.co.uk

Browse more oneBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: oecd.org
Title: artificial [intelligence]({{ ‘intelligence/’ | relative_url }}) job quality and inclusiveness a713d0ad
Link:https://www.oecd.org/en/publications/oecd-employment-outlook-2023_08785bba-en/full-report/artificial-intelligence-job-quality-and-inclusiveness_a713d0ad.html

2. Source: oecd.org
Link:https://www.oecd.org/en/topics/sub-issues/ai-and-work.html

3. Source: oecd.org
Link:https://www.oecd.org/en/topics/ai-and-work.html

4. Source: aubreydaniels.com
Title: is deskilling a threat to safety in your workplace
Link:https://www.aubreydaniels.com/blog/is-deskilling-a-threat-to-safety-in-your-workplace

5. Source: zephoria.medium.com
Title: Deskilling on the Job
Link:https://zephoria.medium.com/deskilling-on-the-job-bbd71a74a435

Published: April 21, 2023

6. Source: oecd.org
Title: algorithmic management in the workplace 287c13c4 en
Link:https://www.oecd.org/en/publications/algorithmic-management-in-the-workplace_287c13c4-en.html

7. Source: oecd.ai
Title: algorithmic management in the workplace
Link:https://oecd.ai/en/ai-publications/algorithmic-management-in-the-workplace

8. Source: ifow.org
Title: making algorithmic management safe for workers new regulation is needed
Link:https://www.ifow.org/news-articles/making-algorithmic-management-safe-for-workers-new-regulation-is-needed

9. Source: oecd.ai
Link:https://oecd.ai/en/working-group-future-of-work

10. Source: oecd.org
Title: making ai work why investing in skills matters
Link:https://www.oecd.org/en/blogs/2026/01/making-ai-work-why-investing-in-skills-matters.html

11. Source: oecd.ai
Link:https://oecd.ai/en/community/gpai-working-group-on-the-future-of-work

12. Source: dl.acm.org
Title: This, in turn
Link:https://dl.acm.org/doi/10.1145/3613904.3642907

Published: May 2024

13. Source: oecd.org
Link:https://www.oecd.org/en.html

14. Source: oecd.org
Link:https://www.oecd.org/en/topics/future-of-work.html

15. Source: oecd.org
Title: full report
Link:https://www.oecd.org/en/publications/how-widespread-is-algorithmic-management-in-workplaces_cda7a114-en/full-report.html

16. Source: oecd.org
Link:https://www.oecd.org/en/topics/policy-issues/future-of-work.html

17. Source: oecd.org
Title: 59e89d7f en
Link:https://www.oecd.org/content/dam/oecd/en/publications/reports/2023/03/defining-and-classifying-ai-in-the-workplace_35d2d069/59e89d7f-en.pdf

18. Source: publications.jrc.ec.europa.eu
Title: JRC Publications Electronic Monitoring and Surveillance in the Workplace
Link:https://publications.jrc.ec.europa.eu/repository/bitstream/JRC125716/jrc125716_electronic_monitoring_and_surveillance_in_the_workplace_final.pdf

Published: November 22, 2021

19. Source: nelp.org
Link:https://www.nelp.org/new-report-details-employers-harmful-use-of-digital-surveillance-and-automated-decision-systems-highlights-urgent-need-for-policy-action/

Additional References

20. Source: businessinsider.com
Link:https://www.businessinsider.com/ai-tools-are-deskilling-workers-philosophy-professor

21. Source: equitablegrowth.org
Link:https://equitablegrowth.org/research-paper/estimating-the-prevalence-of-automated-management-and-surveillance-technologies-at-work-and-their-impact-on-workers-well-being/

22. Source: tuac.org
Link:https://tuac.org/news/new-oecd-report-on-algorithmic-management-reveals-urgent-need-for-worker-protections/

23. Source: uclalawreview.org
Link:https://www.uclalawreview.org/the-consequences-of-automating-and-deskilling-the-police/

24. Source: GOV.UK
Title: assessment of ai capabilities and the impact on the uk labour market
Link:https://www.gov.uk/government/publications/assessment-of-ai-capabilities-and-the-impact-on-the-uk-labour-market/assessment-of-ai-capabilities-and-the-impact-on-the-uk-labour-market

25. Source: sciencedirect.com
Link:https://www.sciencedirect.com/science/article/pii/S0925753525000888

26. Source: onlabor.org
Title: Training Your Replacement, One Keystroke at a Time2 days ago —
Link:https://onlabor.org/training-your-replacement-one-keystroke-at-a-time/

27. Source: academy.evalcommunity.com
Title: how do oecd ai principles support monitoring and evaluation me
Link:https://academy.evalcommunity.com/how-do-oecd-ai-principles-support-monitoring-and-evaluation-me/

28. Source: inc.com
Title: how ai automation is quietly deskilling white collar workers
Link:https://www.inc.com/andrea-olson/how-ai-automation-is-quietly-deskilling-white-collar-workers/91316067

29. Source: researchgate.net
Title: 376490098 Job Security in the Artificial Intelligence and Automation Era
Link:https://www.researchgate.net/publication/376490098_Job_Security_in_the_Artificial_Intelligence_and_Automation_Era