AI Won’t Fix the NHS. People Might - Ideas in Progress: Meaning, Mischief & Mayhem

I had some time in the middle of my working day yesterday and found myself in a freewheeling conversation with a colleague. He said something I’ve heard many times before:

“AI could solve a lot of the NHS’s problems.”

Almost on cue, I got an email from a manager who had been to some conferences. He was very impressed by the “digital strategies,” especially around automation.

These two moments collided in my head — and raised a very real question:

Is technology the panacea for the NHS’s problems?

It’s timely to ask, given the recent 10-year plan from NHS England, which places heavy emphasis on AI and the NHS App.

So let’s unpick this.

The Real Problem: Access, Not Care

I work on the frontlines. I don’t think there’s any doubt that once a patient is in the system, they receive good care. But getting in — that’s the nightmare.

Patients wait 12 hours in A&E
18 months for elective surgeries
8 AM phone queues for GP appointments
A hospital visit for a basic X-ray

Why?

Because they can’t get a GP slot
Because specialists can’t admit directly to wards
Because referral pathways are a maze of dead ends

That’s not a tech problem. That’s a workflow and access problem.

The Myth of NHS Data

People love to say, “AI just needs data.” Sure. But what data are we talking about?

NHS data is:

Fragmented across hundreds of systems
A mix of PDFs, dictated notes, scanned letters, free text, and images
Trapped in outdated infrastructure, some from the 1990s
Not standardised, not labelled, and barely interoperable

Each Trust uses its own stack, and information governance is a fortress — as it should be.

Yes, the NHS holds vast amounts of data. But most of it is:

Unusable for AI.

What the NHS Actually Needs

It’s not AI. It’s clarity, simplicity, and joined-up thinking.

Here’s what I used in a typical DGH job:

One system to view blood/radiology results
Another to access legacy records (ECGs, old notes)
A third to prescribe chemotherapy
A fourth to report blood films
BlueTeq for cancer drugs
A separate one for pregnancy prevention programme forms
And let’s not forget: endless emails between consultants, nurses, admin, pharmacy, and day units to coordinate actual care

In my current role, I’m marginally luckier — chemotherapy prescribing has been integrated into the EPR. But I still need five different logins to treat one patient.

The problem isn’t a lack of tech.
It’s too much fragmented tech, layered onto broken workflows.

The AI Fantasy vs NHS Reality

The NHS plan says:

“We will use AI to reduce waitlists, optimise pathways, predict deterioration, automate notes, and improve equity.”

Sounds great. But here’s the truth:

The problem isn’t triage inefficiency — it’s referrals lost in a black hole
The problem isn’t diagnostic variance — it’s too few staff, too little continuity
The problem isn’t lack of data — it’s bureaucratic blindness

And no — hiring more staff won’t fix it if everyone spends more time on admin than clinical care.

Tech Is an Amplifier — Not a Saviour

When you apply technology to a focused problem with clear workflows, it becomes a force multiplier.

But when you layer tech onto dysfunction, it scales the dysfunction.
It adds chaos, complexity, and confusion.

And when it all crashes?

Tech gets blamed. AI gets blamed.
The real problem — the broken system — stays untouched.

Let’s be honest — most of these AI strategies won’t meaningfully improve frontline care.
They’ll generate a few flashy pilots, some slick dashboards, and a dozen jobs in consulting firms.

Meanwhile, the day-to-day problems — delayed discharges, lost referrals, disjointed records — remain exactly where they were.

So no, the NHS doesn’t need AI.
It needs fewer systems, smarter workflows, and people who can make decisions.

Because AI doesn’t solve problems.
People do.