Will AI replace us? Is the white-collar career promise failing?
Generations have been promised that a University education, a decent suit and a willing attitude would land you a job and a decent white-collar career. Lattes and cheese plants aplenty.
Although the proportion of young people with a degree is now 60%, the UK is seeing rising unemployment rates, especially among young people. Finding a job in the current job market can be endlessly dissatisfying with hundreds of applications and no response.
I work with SMEs in Hackney, East London. How is all the adoption of AI going to affect entry level jobs? And work in general? What can we learn from previous patterns of automation?
In the UK, proposed mass dismissal programs rose by 5% in 2024/25, involving over 267,000 proposed redundancies. UK businesses accelerate layoffs amid AI surge and legislative uncertainty. Full adoption of generative AI could save nearly 24% of private-sector workforce time, equivalent to the output of 6 million workers. Predictions are that nearly 40% of global employment will be affected by the end of the decade. The Impact of AI on the Labour Market
The Centre for AI and Digital Policy (CAIDP), a US non-profit research organisation, produces its report on global AI policy every year. Here’s a summary of the CAIDP’s 2025 Index on which job roles will benefit from AI automation and which will suffer. The word clerks keeps coming up – lower-skilled, administrative roles i.e. the first step on the corporate ladder. CAIDP Index 2025 – Center for AI and Digital Policy
| Fastest-Growing Jobs (Technological Focus) | Fastest-Declining Jobs (Clerical/Administrative) |
| Big Data Specialists | Postal Service Clerks |
| FinTech Engineers | Bank Tellers and Related Clerks |
| AI and Machine Learning Specialists | Data Entry Clerks |
| Software and Applications Developers | Cashiers and Ticket Clerks |
| Security Management Specialists | Administrative and Executive Secretaries |
| Data Warehousing Specialists | Printing and Related Trades Workers |
| UI and UX Designers | Accounting, Bookkeeping, and Payroll Clerks |
| Renewable Energy Engineers | Telemarketers |
Imagine it’s the early noughties. The introduction of self-service tills, along with the promise of no more queuing. You would now be in control of your own shopping experience, could scan and pack your own bags at your own pace, and do so privately without feeling undue pressure. Based on the estimates at the time, what would cost the supermarket £2.50 in staff costs would now cost them 15 pence, with the implication that prices would also be affected. You’d soon be waiting less and spending more for your weekly shop.
Even the staff are about to be liberated from monotonous and repetitive tasks. They will take on more supervisory roles and had more flexible interactions with customers.
A 2003 Nielsen survey found that 52% of shoppers considered self checkout lanes to be “okay,” while 16% said they were “frustrating.” Thirty-two percent of shoppers called them “great.” Nobody likes self-checkout. Here’s why it’s everywhere | CNN Business
“It hasn’t delivered anything that it promises …. Stores saw this as the next frontier… If they could get the consumer to think that [self-checkout] was a preferable way to shop, then they could cut labour costs. But they’re finding that people need help doing it, or that they’ll steal stuff. They ended up realising that they’re not saving money, they’re losing money.” Christopher Andrews, Associate Professor and Chair of Sociology at Drew University, USA.
A 2025 Serve Legal study found 37% of shoppers admit to stealing at self‑checkouts, contributing to retail crime losses of £4.2bn annually. As supermarkets removed manned tills and pushed self‑checkout, more shoppers reported waits up to three minutes or more instead of no queue.
Studies of over 6,000 workers cited by The Grocer describe high stress, abuse from frustrated customers, and a sense that front‑end staffing levels are no longer sustainable. Proof that self-checkouts are causing a decline in customer service | Comment & Opinion | The Grocer
Only 10% of customers (from a 300 UK customer sample size) want more automation in store (research conducted by GetApp in February 2024) UK survey shows self-checkouts top store automation technologies
Supermarkets typically operate on a 3% profit margin. After billions of pounds in investment in the technology, the promised savings of replacing labour with automated tills are not there. Include shoplifting, security and reputational effects and the margins are tight but, crucially, there are there. There is still a margin of profit.
As competitors have all installed the technology to seem equally interested in their customers, there’s not really any alternative so, even after a poor experience, you can’t vote with your feet and shop somewhere that hasn’t converted to self-service.
Stores today are catering to shoppers who perceive self-checkout to be faster than traditional cashiers, even though there’s little evidence to support that. But, because customers are doing the work, rather than waiting in line, the experience can feel like it’s moving more quickly. Nobody likes self-checkout. Here’s why it’s everywhere | CNN Business
According to the GetApp survey mentioned earlier, consumers are now positive about self-checkout tills:
UK survey shows self-checkouts top store automation technologies
| Industry activity | Public perception | |
| Step 1 | Automation is driven by cost savings (primarily labour) and associated profit margins | Change is presented as progressive with promises of improved experience |
| Step 2 | Accelerated adoption of automation by a competitive market | Poor experience and sympathy towards job losses and the marginalised minority (See Margins and Marginalisation below) |
| Step 3 | Accustomisation as experience improves beyond a lower threshhold of expectation | Customer perception of independence develops as they get more skilled at the delegated work |
| Step 4 | The automated process is normalised and issues minimalised | Sympathy for job losses and the marginalised minority evaporates |
Of course, not everyone has accepted the automation of everyday life tasks. Consider how automation can instead become exclusion. In the 10 years since 2013/14, the estimated number of disabled people in the UK has increased by 4.9 million. Among 16 to 24 year olds, there are twice as many reporting as disabled, increasing from 8% to 18%. UK disability statistics: Prevalence and life experiences – House of Commons Library
Imagine you want to use a Self-Service checkout and you have a visual impairment. You have to use a touchscreen, there is no headphone jack for headphones to have a voice explain what you can see, the flat screen does not provide any tactile means to identify where you should press. Or you have a motor impairment. Cramped self-service areas. Wheelchair users cannot reach up and forwards enough, scanning requires precise movements, machines freeze if there’s a delay.
“I have MS and am in a wheelchair, and find it almost impossible to use the automated tills. I find it hard to juggle everything on my lap — purse, handbag, unwieldy basket, various cards — and very difficult indeed to scan and move the shopping fast enough. The first time I tried to do it, the assistant was busy and I sat there, holding everyone else up, waiting for her to help. I find these machines very upsetting because effectively they stop me from going shopping.” Linda Riordan, 60
“I have Parkinson’s disease and loathe automated tills. They are demoralising and make me feel stupid. Parkinson’s affects everyone who has it in different ways, but very often the tremor gets worse if you feel stressed or upset … I only ever use them when I’m absolutely forced to. I would rather queue halfway around the shop for a manned checkout.” Lin Merchant, 64
In 2024, several major UK and US retailers began reintroducing manned tills specifically because of feedback regarding accessibility and the “unfriendly” nature of automated lanes for vulnerable customers.
So, where are we on the AI automation of white-collar work?
Delving deeper into the CAIDP’s 2025 Index, we see the expectation of a gentler transition than the headlines might suggest. Enter the Jevons Paradox. This states that as a service or resource becomes more efficient, the overall demand for what it creates rises.
LED lights are more efficient. Older lights are replaced but new LED lights displays on signs or other media have been introduced. The fact they’re cheaper means there are more of them.
AI makes coding and software creation significantly cheaper and faster. What happens? More software, more tailored offers and more use cases. And this creates new specialist roles. The transition is less sudden and cliff-edge and more gradual as companies learn the skills to use AI competently and humans are employed to manage the areas it can’t handle or needs supervision for.
“It’s certainly the case right now… that the human-AI combination, the ‘cyborg,’ is superior to AI alone.” — David Oks, Essayist
Human institutions are slow to adapt and change to new technologies. I interviewed a Finance Director in a niche sector several years ago and he told me there was still a niche part of taxation where the form had to be sent by fax. They still had a working fax machine. Yes, AI is faster in its lane but it can’t comprehend the distinct personality of a company’s management structure and how Brian likes his folders labelled.
Step 1.
AI is a new wave of automation. It promises to do the repetitive, resource-heavy tasks you hate for you and free you up for more satisfying work.
We will see an acceleration of adoption among the companies who can afford large-scale AI investment. Recruitment among big companies will slow or freeze. However, initial hopes of significant profit leaps will not materialise.
At the leading edge, Klarna (the digital bank with the buy-now-pay-later instalments feature)have claimed its AI customer assistant does the work of 853 full-time agents and this saves them $60 million a year. However, overall customer service and operations costs were slightly more year-over-year ($50M vs $42M in Q3 2025). Klarna Investor Relations, “Q4 2025 Results & Strategic Update,” Feb 19, 2026. They have halved their workforce since 2022 as their revenue has doubled. Klarna Group plc – Klarna Accelerates U.S. Growth and Delivers $1bn Revenue Driven by Rapid Banking Service Adoption
However, cost savings were masking an underlying issue. A quarter of the customers with resolved issues were coming back. Their problems still existed after all. The company have now pivoted as of February 2026 to a model that re-introduces a human for more involved or high value queries.
Klarna Saved $60 Million with AI. Then It Had to Rehire the Humans.
In May 2025, in a Bloomberg interview, Klarna CEO Sebastian Siemiatkowski admitted: “As cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality … Really investing in the quality of the human support is the way of the future for us”.
Step 2.
It seems to me we’re still at early Step 2.
For most people in most work or office situations, AI certainly helps but the quality of the results is poorer than expected. Over-reliance on what AI can be trusted to do causes issues down the line.
Partly, this is because we haven’t been trained how to use it well enough and partly because the AI models aren’t reliable enough.
The AI automation of repetitive administrative work is not a wholesale replacement of white-collar clerical workers. At least not yet. Instead, it’s a way to build efficiency into the admin work being done so that employees can achieve and produce more with less human resource.
Traditional ways of working take time to adapt and add blockers even when companies do want to automate. People don’t trust new technologies. The change will be more gradual and will probably a decade or two.
What it does mean is that new clerical roles will stall. The need for new entry level and junior roles is where the change is taking place.
However, new roles in quality assurance, AI agent management and software development will also be created.
SMEs with specialist skills can fill the gap that a drop in entry level roles will leave behind. Tailored software solutions or management services will offer businesses alternatives to developing new staff. This out-sourcing of talent and responsibility will become more attractive as the risks of AI become more apparent. But that’s a future blog.
If you want to find Clarity and enjoy research-based customer insights, Clarity offers in person customer interviews for SMEs in Hackney. We offer a package of 5 half hour interviews per month and promise 5 actionable insights from each interview, providing you with authentic, independent customer feedback and a suite of marketing materials to build your reputation and word of mouth referrals.
We've all heard of zero-hour contracts (ZHCs) and usually with some kind of negative spin.…
SMEs make up 99% of all UK businesses and employ 16 million people but they…
On 26th November, the long-discussed Budget dropped and delivered something very different for SMEs and…
Did you know that your SME could be granted or even win money? Money that…
How do I boost my business growth? Would a cash injection help? Yes, the big…
Nvidia becoming the richest business in the world, Open AI turning over billions of dollars…