“A graduate trainee does what a senior associate did three to five years ago.”

That’s Phillippa O’Connor, PwC UK’s chief people officer, explaining their workforce transformation to the Financial Times this week. It’s also a perfect summary of what I called the “experience paradox” when I wrote about the end of mid-tier knowledge worker careers nearly a year ago.

The article reveals something striking: PwC UK received 60,000 applications for 2,000 entry-level positions last year, a 35% increase as graduates hunt for roles in a white-collar jobs market being reshaped by AI. At the same time, the firm has abandoned plans to add 100,000 to its worldwide headcount, frozen starting salaries and cut “a couple of hundred” positions from recent graduate intakes.

This isn’t news to anyone who’s been paying attention. It’s confirmation that the workforce transformation I predicted is already complete.

For full disclosure, I led a content team for a digital project at PwC years ago. There was a constant influx of eager graduates with clipboards. It was a ‘sink or swim’ environment. Most people were generalists, with external experts – like me – brought in and milked for every last bit of knowledge on a subject. Everything had to be turned into shared knowledge, often PowerPoint presentations, from the most basic to advanced strategies. Everyone was considered replaceable if they either didn’t hit their KPIs or had no more knowledge to share. But at least there were free pizzas every Thursday, so that’s okay then.

What I said in March 2025

Here’s what I wrote ten months ago about the mid-tier knowledge gap:

“A partner at a London law firm told me recently that they’re completely rethinking their staffing model. ‘We used to need a pyramid of associates,’ she explained. ‘Now we’re moving toward what looks more like an hourglass: fewer mid-level positions with more senior strategic advisors and junior AI wranglers.'”

I called it the “great middle squeeze”, the recognition that AI tools were compressing the value gap between junior and mid-tier professionals while leaving genuine expertise at the top still valuable. The traditional career arc of gradual knowledge accumulation was being disrupted as AI systems captured significant portions of that accumulated knowledge.

I also predicted the acceleration timeline: “Mid-tier knowledge workers don’t have the luxury of a decades-long adjustment period. The foundations of their professional value are being reconfigured in real time.”

That was March 2025. By February 2026, PwC is describing it as operational reality.

What PwC is admitting now

Let me pull out the key admissions from O’Connor’s interview:

The hourglass structure exists: “There’s a version of that pyramid where you take out a bigger component of the bottom of the pyramid.”

The experience paradox is real: “A graduate trainee does what a senior associate did three to five years ago” because of AI productivity gains.

The career pipeline is broken: They’re receiving 35% more applications while hiring hundreds fewer positions and abandoning global headcount targets.

The middle is being eliminated: Greater weight on “human skills” at entry level because AI handles what mid-tier professionals used to do.

But here’s where it gets interesting: O’Connor frames this as strategic choice rather than inevitable consequence. She says PwC has “chosen not to replace or offshore some routine work” because they want junior staff to develop judgment.

This is corporate double-speak at its finest.

The structural contradiction

PwC wants you to believe two incompatible things simultaneously:

1. They’re protecting junior positions so graduates can develop judgement that “could look black and white to an AI machine”.
2. Those same graduates are already performing senior associate work through AI assistance.

If point 2 is true – and O’Connor says it is – then point 1 is theatre. You can’t claim you’re preserving learning opportunities while admitting the learning curve has been compressed out of existence.

The numbers expose the lie. They say they value developing judgement, but they’ve:

• Cut hundreds of entry positions
• Frozen starting salaries
• Abandoned headcount growth targets
• Increased applications-per-position by 35%
• Admitted graduates now do work that used to require years of experience

This isn’t preserving career development. This is managed decline with a PR strategy.

And here’s the bit that should terrify those 60,000 applicants: O’Connor says they’re now looking for candidates with “communication, empathy, emotional resilience and the ability to handle change” because AI handles the technical work.

Translation: we need people who can cope with doing senior-level work without senior-level experience, support or career progression, because the middle rungs we used to provide have been removed.

The expertise crisis nobody’s talking about

This creates a problem that goes far beyond individual career disruption. Let me explain using an example from my own work.

When I was helping design Universal Credit application processes, the most valuable people on the team weren’t the senior strategists or the junior researchers. They were the mid-tier content designers who’d spent years seeing how stressed claimants actually interacted with our systems.

They had the scarring, the accumulated knowledge of what goes wrong that only comes from repeated exposure to failure. They knew which questions triggered panic, which terminology confused people under cognitive load and which design patterns failed for users in crisis.

That knowledge wasn’t in anyone’s training manual. It couldn’t be extracted by asking them what they knew. It was embodied expertise developed through years of mid-level work.

Now imagine that role being eliminated. The graduate uses AI to generate content based on best practices. The senior strategist reviews the output against policy requirements. Both miss the detail that the mid-tier professional would have caught: this wording will devastate someone who’s just been bereaved.

PwC is creating exactly this expertise vacuum in audit, tax and consulting. When graduates use AI to build financial models without years of experience seeing how models fail, who catches the errors? The senior partner is too far from the detail. The AI has no judgement. The mid-tier professional who would have spotted the problem no longer exists.

O’Connor acknowledges this indirectly when she says they need juniors to learn “subjectivity” in auditing, to see nuance that “could look black and white to an AI machine”. But she’s describing a learning process that requires time, supervision and progression through mid-tier roles they’re simultaneously eliminating.

You can’t compress judgement development the way you can compress task completion.

Why this matters for vulnerable users

Here’s where this connects to my broader work on trauma-informed content and Consumer Duty compliance.

The Financial Conduct Authority fined firms billions for failing vulnerable customers. The core failure? Organisations treated vulnerable customers as an edge case rather than recognising vulnerability as a universal state under specific conditions.

The same pattern applies to workforce transformation. PwC treats the expertise crisis as an edge case, something to mention in passing while focusing on productivity gains. But expertise isn’t an edge case. It’s the foundation of professional judgement.

When you eliminate mid-tier professionals who developed judgement through experience, you’re not just creating career disruption. You’re systematically removing the people who prevent catastrophic failures for users in vulnerable states.

I’ve seen this pattern across four decades of government digital transformation. The worst failures – the benefit systems that drive people to suicide, the services that exclude elderly users, the processes that traumatise bereaved families – invariably involve junior people implementing “best practice” without the scarring to know when best practice fails.

The mid-tier professional would have caught it. But organisations decided that layer was “inefficient”.

Where’s the accountability?

If PwC were selling a financial product that worked like their current career structure – “invest in this degree, we’re hiring for these positions, oh actually we’ve eliminated most of them but don’t worry we’re choosing not to eliminate them too fast” – the FCA would fine them for misleading customers.

But workforce policy gets a pass on the same structural dishonesty.

Those 60,000 applicants invested years and substantial resources in qualifications based on career promises that no longer hold. They were told: get a good degree, join a professional services firm, develop expertise, build a career. Now they’re being told: actually, AI does most of what you thought you’d spend 5-15 years learning, but we need you to have emotional resilience about it.

Where’s the accountability for that bait-and-switch?

O’Connor says the increase in applications has allowed PwC to “raise the bar” and select candidates able to deliver higher value work. This is extractive logic wrapped in meritocracy language. They’ve made the competition more brutal while offering fewer positions with less career progression and they’re framing it as an opportunity.

The workers aren’t the problem requiring a tech solution. The workers ARE the economy. When you systematically eliminate the career paths that create expertise, you’re not optimising; you’re destroying the foundations of professional judgement while increasing the precarity of those who remain.

What happens next

I’m not surprised by PwC’s admission. I predicted this pattern nearly a year ago and I’ve been seeing it play out across every knowledge work profession I study.

What surprises me is the brazenness. Ten years ago, firms would have been defensive about eliminating career progression. Now they admit it openly while claiming it’s a strategic choice.

Here’s what I think happens next:

Short term (1-2 years): More firms follow PwC’s lead in cutting entry and mid-tier positions while maintaining the fiction that they’re “developing talent differently”. Graduate unemployment rises. Starting salaries stay frozen or decline. Competition for remaining positions intensifies.

Medium term (3-5 years): The expertise crisis becomes visible. Organisations that eliminated mid-tier professionals start experiencing catastrophic failures they can’t diagnose because the people who would have caught them are gone. Regulators start asking questions about risk management when your entire workforce is either junior or executive.

Long term (5-10 years): Either organisations recognise that you can’t compress judgement development and rebuild career structures or we accept a permanent two-tier system where most knowledge workers never develop genuine expertise. The economic and social consequences of that second option are grim.

For those 60,000 applicants chasing 2,000 positions: I’m genuinely sorry. You were promised a career structure that was already being dismantled. The advice you were given – get qualified, work hard, develop expertise – wasn’t wrong. But the organisations you’re applying to changed the deal without telling you.

For PwC and similar firms: you’re creating an expertise vacuum you don’t fully understand yet. The productivity gains look impressive until you start experiencing the failures that only mid-tier professionals with years of scarring would have prevented. And you will experience those failures.

For regulators and policymakers: this is structural exclusion wrapped in productivity metrics. If you apply the same accountability standards to workforce transformation that you apply to financial products, the bait-and-switch becomes obvious. Start asking the hard questions before the expertise crisis becomes catastrophic.

As for me: I predicted this in March 2025. PwC confirmed it in February 2026. The hourglass workforce is already here. The question now isn’t whether it’s happening, but whether we’re going to hold anyone accountable for the consequences.