AI job losses and what history tells us might happen

Summary

• The speed of technological change often outpaces industry and organisations' ability to adapt, causing concern — AI is no different
• While new technologies tend to threaten traditional roles, they also create a slew of new ones
• Workers must be involved in how AI is used to redesign their roles in order to enhance day-to-day work rather than undermine it
• AI implementation must be undertaken with thought for human wellbeing at its core
• The key to whether a technology like AI is destructive or liberating lies with how society chooses to use and interact with it

It’s an irony, or maybe it’s a contradiction: when you ask most people to describe AI — what it is and how it works — they'll probably start with a chatbot or generative AI. They’ll likely talk about AI that builds and makes things: words, pictures, video and sounds, often seemingly from nothing. They’ll talk about and describe a productive and creative form of AI.

But then, ask those people what worries them about AI, and they’ll likely describe fear about what the technology means for jobs, their lives and livelihoods. AI is, in a way, a potentially destructive technology too.

Every new technological revolution, of course, arrives with a mixture of promise and dread. In that regard, artificial intelligence is no different from the new technologies that preceded it.

Today, the conversation is dominated by fears that AI will hollow out entire professions, deskill workers and concentrate power in the hands of those who own the technology. Yet these anxieties about new technologies are not new. They echo, almost word for word, the debates that surrounded earlier waves of innovation — from the arrival of silicon chips to the spread of word processing in the office.

To explore these echoes, we’ve taken a trip back through BCS’ archives and looked at two specific articles: Jonathan Sleigh’s Chips (The Computer Bulletin, June 1979) and Jobs and A. Wight’s Word Processing–A Trade Union View (The Computer Bulletin, March 1980)

We can trace recurring patterns in how societies respond to disruptive technologies. More importantly, we can draw lessons that might help us navigate the AI transition with greater clarity and fairness.

The speed of change and the shock to society

One of the most striking parallels between the rise of AI and the earlier revolutions is the rapid pace at which new technologies emerge. Sleigh’s account of the silicon chip era emphasises the ‘astonishing speed’ with which microelectronics reshaped industry. There was a sense that society was being overtaken by forces it barely understood. The pace of change outstripped the ability of institutions — governments, unions, employers — to adapt.

Wight’s trade union perspective on word processing echoes this same urgency. Apex, the now-defunct Association of Professional, Executive, Clerical and Computer Staff, found itself needing to ‘urgently assess the prospective novel threats and opportunities’ posed by word processing. The technology was spreading faster than the frameworks needed to manage it. Employers, for their part, often reduced the conversation to a crude equation, very much reflecting the prejudices of the time: ‘one terminal in, one girl out.’

This mismatch between technological acceleration and institutional preparedness is a recurring theme. AI today is advancing at a pace that makes even the silicon‑chip revolution look leisurely. The lesson from BCS’ historical accounts is clear: when technology moves faster than society’s ability to absorb it, fear fills the gap. That fear often manifests as predictions of mass unemployment, social unrest and the erosion of meaningful work. Whether those predictions come true could depend less on the technology itself and more on how we choose to respond.

Job destruction or job transformation?

Both Sleigh and Wight grapple with the central question that now dominates the AI debate: will new technologies destroy jobs, or change them?

Wight’s analysis of word processing is particularly nuanced. He acknowledges that the technology could be used to cut costs by eliminating typists and secretaries. But he also argues that it could ‘provide a splendid opportunity to change the content of office jobs into challenging and satisfying work.’ The problem, he suggests, is not the technology but the way employers choose to deploy it. Many saw only the possibility of small savings — ‘typically no more than 2-4% of office labour costs’ — and missed the larger opportunity to improve support for managers and professionals.

Sleigh’s discussion of silicon chips follows a similar pattern. Chips threatened traditional roles, but they also created entirely new industries and skill sets. The challenge was ensuring that workers could transition into these new roles rather than being left behind.

The parallel with AI is clear. AI can automate tasks but also augment human capabilities. It can deskill work, but it can also elevate it. The determining factor is not the algorithm but the intention behind its use. If AI is deployed primarily as a cost‑cutting tool, job destruction becomes a self‑fulfilling prophecy. If it is used to redesign work, enhance productivity and create new forms of value, the outcome can be far more positive.

Economic context matters

Wight makes a point often missing from modern debates about technology: the broader economic climate determines how disruptive innovation might become. He highlights that between 1968 and 1978, total employment in the UK rose by only 100,000 — a decade of almost no job growth.

At the same time, the labour force was projected to expand by a further two million by 1991. In other words, far more people would soon be looking for work, while the economy had already shown it could not create enough jobs to absorb them. In such a situation, even small reductions in employment caused by new technologies could have serious social consequences. Wight warns that large scale unemployment risks triggering ‘serious unrest’ and pointedly asks, ‘How many millions have to be unemployed before there is a revolution?’

The lesson for AI is straightforward: technology does not operate in a vacuum. Its impact is amplified or softened by demographic trends, economic growth, labour‑market flexibility, and social safety nets. AI introduced during a period of economic expansion will have a very different effect from AI introduced during stagnation or contraction. Understanding this context is essential if we are to avoid repeating past mistakes.

The importance of worker involvement

One of the strongest themes in Wight’s article is the argument that workers must be involved in the design and implementation of new technologies. Without such involvement, he warns, computer‑based systems ‘only serve to make interesting jobs boring, and the boring jobs even more monotonous.’ With involvement, however, technology becomes an opportunity to redesign jobs to better suit the ‘varied needs and capabilities’ of staff.

For you

Be part of something bigger, join BCS, The Chartered Institute for IT.

Apex advocated for ‘comprehensive New Technology Agreements’ to ensure technological change occurred ‘by agreement’ rather than by imposition. This approach was not merely protective; it was constructive. Workers understood the realities of their jobs better than anyone else, and their insights could improve system design, increase acceptance and boost motivation.

Sleigh’s reflections on the silicon‑chip era reinforce this point. When workers were excluded from decision making, resistance and fear grew. When they were included, transitions were smoother and more productive.

The relevance to AI could not be clearer. AI systems that are imposed on workers without consultation risk creating stress, resentment, and unintended consequences. AI systems developed with worker input are far more likely to enhance rather than undermine human work.

The human cost of poorly implemented technology

Wight devotes significant attention to the health and safety implications of early word‑processing terminals. While fears about radiation proved ‘largely imaginary,’ the real problems — eyestrain, backache, stress — were widespread. Most terminals, he writes, were ‘an ergonomic disaster,’ installed without regard for the working environment.

This is a powerful reminder that technological revolutions are not just economic or organisational events; they are human experiences. Poorly implemented technology can cause physical and psychological harm. Sleigh’s discussion of microelectronics similarly highlights concerns about stress and deskilling.

AI may not strain the eyes or the back, but it introduces new forms of pressure: constant monitoring, accelerated workflows, reduced autonomy and the ever‑present fear of redundancy. These are not abstract concerns; they are lived realities for many workers. The lesson from history is that such harms are avoidable if technology is implemented thoughtfully and with genuine concern for human wellbeing.

Conclusion

The fears surrounding AI are not new. They echo the anxieties that accompanied the rise of silicon chips and word processing. In both cases, the technology itself was neither inherently destructive nor inherently liberating. The decisive factor was how society chose to use it. If we treat artificial intelligence as a tool for cost‑cutting, we will repeat the mistakes of the past. If we treat it as an opportunity to redesign work, enhance human capability and share the benefits of progress equitably, the outcome can be far more hopeful. History does not give us a blueprint, but it does give us a warning— and a chance to do better this time.

After close reading of the original articles, AI (Copilot) was prompted to find and draw parallels between the two cited articles and to structure the above article. The article was edited by a BCS editor.