The Replacement Myth

I get asked three questions quite consistently about AI, and they all focus on one theme — “Will AI replace us?”

The three questions are:

1. Which LLM is the best for the type of work I’m doing?

2. What AI course do you recommend?

3. Which jobs will be replaced by AI in the next 12 months?

That last question is the one where I feel the most anxiety and tension emanating from the person asking it, and I understand why.

Since ChatGPT launched in November 2022, entry-level job postings in the UK have fallen by 32%. Graduate roles, apprenticeships, junior positions — collectively they dropped from 28.9% of the UK job market to 25% by May 2025, the lowest since the pandemic.

A King’s College London study found that firms with high AI exposure reduced junior positions by 5.8% on average. And according to the ONS, nearly one million 16-24 year olds in the UK are now NEET — not in education, employment or training — a figure that is rising, and close to its highest level since 2014.

Those numbers are real and the anxiety is warranted.

AI’s impact on jobs

There is no escaping the reality that AI will replace certain types of jobs by 2030, if not before. This pattern is no different from the one that played out as industry moved from the Agricultural era to the Industrial era, and then to the Digital era. Jobs become obsolete as technology progresses. There is no escaping this. There is no resisting it. There is only accepting it.

Or there is planning for it.

A Harvard Business School study of job postings following the launch of ChatGPT found that postings for automation-prone roles — those involving structured, repetitive tasks — fell by 17% per quarter per firm. Roles like correspondence clerks, medical transcriptionists, and junior administrative positions saw the sharpest declines.

These are not abstract statistics, they are the entry points that previous generations used to build careers.

The question nobody is asking correctly

When I’m asked “which jobs will be replaced by AI in the next 12 months?”, I encourage the individual to reframe their thinking. Rather than focusing on which jobs are disappearing, ask instead: which jobs are being augmented?

The same Harvard study that recorded a 17% decline in automation-prone job postings also recorded a 22% rise in postings for augmentation-prone roles in the same period. Jobs that require more analytical, technical, and creative work — roles where AI enhances rather than replaces the human doing them.

That 22% rise does not cancel out the 32% drop in entry-level postings. I want to be clear about that. But it does point to a direction of travel that matters.

What augmented actually means

Here is where I think the conversation goes wrong.

When people hear “augmented jobs”, they often picture highly specialised roles — engineers, data scientists, clinical specialists. And yes, those roles are growing. But that framing creates a trap. It implies that augmentation is something that happens to certain categories of skilled people, while everyone else waits to be replaced.

That is not what is happening.

Augmentation is not about which job you hold. It is about which part of the job you are doing. And this is the shift that matters.

The repetitive, structured, rule-following parts of most jobs — the data entry, the scheduling, the drafting of routine correspondence — those are the tasks being absorbed by AI. What is left is not a lesser version of the job. It is the part that required a human to begin with.

The judgement call the admin assistant makes when a request is ambiguous. The instinct the HR professional uses when a policy does not quite cover a situation. The relationship a customer service manager holds with a difficult client. The pattern recognition a junior analyst develops over months of working close to data.

None of that disappears when AI removes the repetition around it. In fact, without the repetition filling the day, that capability becomes the job.

McKinsey estimates that more than 70% of today’s skills are applicable in both automatable and non-automatable work. The value of those skills is not diminishing — but the work context they operate in is changing quickly.

I just want to say; I am aware this argument can sound like cold comfort if you are in a role that has just been cut, or if you are 22 and cannot get your first job because the entry-level positions have gone. That is not an ideological problem. It is a structural one, and it deserves a structural response.

The honest counterargument

The reframe from replacement to augmentation is useful. But it is not the whole story.

32% of organisations surveyed by McKinsey expect workforce decreases as a result of AI. The transition to augmented work is not smooth, and it is not equal. People in lower-skilled, lower-wage roles face a harder journey than those already doing work that requires complex judgement. Young people entering the workforce are competing for a shrinking number of starting points. The skills that employers say they value — critical thinking, creativity, adaptability — take time to develop, and they cannot be developed without employment to develop them in.

This is the structural gap that a reframe alone cannot close.

The UBI conversation

It is why, in February 2026, a UK government minister suggested that universal basic income should be considered as a response to AI-driven job displacement. That is no longer a fringe idea. It is now policy-adjacent.

The argument for: if AI concentrates productivity gains among capital owners and a smaller skilled workforce, a universal income floor ensures that displacement does not translate into sustained poverty for those left behind. Evidence from UBI pilots shows that people do not stop working when given a basic income. Many use it to retrain, start something new, or move from precarious work into more meaningful work.

The argument against: providing a UBI of around £11,000 per person in the UK would require a flat income tax of approximately 45%. Running that alongside the existing welfare system is fiscally difficult at best, untenable at worst. And income support, however well designed, is not the same as purpose, structure, or the accumulated experience that employment builds over time.

UBI is a cushion, not a strategy. It buys time, but it does not answer the harder question, which is what we are actually preparing people to do.

The question for People Leaders

If you lead a People function, the macro debate about UBI is not your lever to pull. But the structural conversation about what work looks like in your organisation — that one is.

The organisations navigating this well are not the ones with the most advanced AI tools. They are the ones asking a different question.

Not: which roles can we automate?

But: which human capabilities are we building, protecting, and creating the conditions for?

Because the version of your workforce that emerges from this period of disruption will be shaped less by which AI tools you adopted, and more by whether you designed work that still had something genuinely human at the centre of it.

The Replacement Myth is the belief that this is binary. That jobs either survive intact or disappear entirely. The more uncomfortable truth is that most jobs are changing, and the people inside them will either be prepared for what they are becoming, or they will not.

That preparation does not happen by accident. It is a leadership decision.