Posted on: 14/10/2026

Thought Leadership

Michelle O’Neill reflects on how her enthusiasm for AI has evolved as she explores both its opportunities and unintended consequences. Through the “Mixtape Years” theme, she considers how AI can improve productivity and transform work, while also creating new complexity, risks and demands. Join this year’s AI Festival to explore these challenges, share ideas and shape what comes next.

Michelle O'Neill Principal Consultant OD, Leadership and Learning at West Midlands Employers

In 2025, I think I said to everyone…..’I’m not an AI expert, but I am an AI enthusiast’. In the latter part of 2026, I think my stance may seem a little more confused. Am I an AI enthusiast?

Yes…..No…..Maybe.

Which, if I'm honest, probably sums up where many of you are right now.

You'll know I've become slightly obsessed with AI. I've used it every day. I've encouraged others to experiment. I've challenged myths, celebrated successes, facilitated conversations and probably mentioned it in far more meetings than people would have liked.

For a while, I loved almost everything about it.

Every week brought another breakthrough.

Every conversation revealed another possibility.

Every demonstration felt a bit like watching the future arrive early.

So, why do I find myself feeling and sharing as many considerations as I do opportunities now? The Mixtape Years, the title of this years AI Festival, is so spot on! That's exactly where I am.

Just like the mixtapes many of us carefully curated back in the 80’s and 90’s, some tracks will become classics and others are quickly being forgotten.

The more I work alongside you, the more I realise we're still figuring this out together. Some organisations are cautiously exploring. Others are redesigning services. Many are somewhere in between. Even within the same organisation people can hold wildly different views at exactly the same moment. That variation feels entirely natural. In many ways, it might even be healthy.

Track One: Why I loved it

Let's start with the obvious bit. AI is impressive. Really impressive.

We've now seen enough examples across local government to move beyond ‘sounds like a good idea’ and into evidence. In work we've supported, officers have used AI to analyse consultation responses, build reports, cleanse data, draft communications, create knowledge repositories and reduce administration. Some activities have reduced from hours to minutes. Others have seen substantial reductions in effort whilst improving consistency and quality.

Increasingly, though, I'm finding that the productivity story isn't the most interesting story. It's just the easiest one to tell.

Track Two: What’s that in my back pocket?

Over the last year I've found myself reading two very different sets of headlines. On one side, there are stories about productivity, efficiency and groundbreaking breakthrough technologies.

On the other, we're beginning to hear more reports of organisations receiving increasingly sophisticated complaints, grievances, challenges and appeals that appear to have been substantially supported by AI.

You can’t deny it, AI is now in everyone’s back pocket. AI is democratising expertise and that changes the game massively. That will change relationships entirely.

Employees can now research policy, legislation and employment practice in seconds.

Citizens can write SEND or housing complaints like quasi-solicitors, escalating complexity without necessarily getting closer to a resolution.

Managers can respond to employees with polished facts and policy, but little empathy or personal connection.

People can construct arguments faster, more thoroughly and with greater confidence than ever before.

None of that is inherently bad. In fact, much of it could be positive.

Better informed citizens and employees should be welcomed. Better equipped managers could be a blessing. But……there may be unintended consequences.

A complaint that might once have arrived as three paragraphs can now arrive as twenty-seven pages. A simple challenge can become a fully referenced dossier. A straightforward question can become a lengthy exchange requiring considerable time, judgement and expertise to untangle.

One of the most interesting ironies may be that councils increasingly need AI to help navigate communications that have been influenced by AI. AI can summarise lengthy submissions, identify themes and surface the real issues beneath the volume.

But the work does not disappear; it might just move elsewhere. The same could happen with the anticipated value or savings we expected – it simply gets absorbed elsewhere.

Track Three: The cognitive choreography problem

Now those who do know me, will have heard one of my favourite phrases of 2025 – ‘cognitive choreography’. Organisations finding their perfect human and tech blend.

I still love it and will continue to use it. But again, experience is telling me it doesn’t quite cut it. The challenge now is orchestrating thousands of enhanced intelligences interacting with one another. That’s one heck of a mixtape!

Citizens have AI.

Employees have AI.

Managers have AI.

Consultants have AI.

Suppliers and contractors have AI.

Trade unions have AI. Councillors have AI.

Everyone is becoming more capable.

Everyone is becoming more informed.

Everyone is becoming more articulate.

And everyone is becoming more productive.

Perhaps the next leadership challenge isn't cognitive choreography, perhaps it's cognitive complexity. How do we lead, communicate, collaborate and make decisions in a world where almost everybody arrives armed with their own AI-powered thinking partner?

The good, the bad and the ugly

The challenge now is not to decide whether AI is simply good or bad. It is to help organisations and their people work out the good, the bad and the ugly quickly enough to make better choices. We need to understand where AI genuinely improves work, where it creates new risks or shifts effort upstream, and where the promise is still running ahead of the reality.

That learning matters because experimentation is only the starting point. Scaling AI requires much more than buying licences and encouraging people to have a go. It means being clear about the work we are trying to improve, redesigning processes around what the technology can now do and building the confidence, judgement and safeguards people need to use it well.

The move towards an agentic era now is pretty much a pre-determined destination. But agents that undertake tasks, interact with systems and consume services as they work bring a very different cost model from the one most organisations understand today. The economics of AI may shift from relatively predictable licence costs to variable consumption, making choices about purpose, governance and value even more important.

As I explore in my regional briefing note Who knew, AI more human than we thought, the more technology enters work, the more deliberate we need to be about the human bit.

Final track

So yes, I do love all the possibilities AI, copilot brings. But I love it differently. I remain a firm believer everyone needs to understand this thing. Despite all the headlines, predictions and opinion pieces, I suspect the organisations that benefit most from AI won't be those with the biggest budgets, the most licences or the most sophisticated technology.

They'll be the ones that create the conditions for people to think, learn and adapt together. They'll recognise that the real challenge isn't simply adopting AI. It's helping humans and technology create value together in a way that works for citizens, colleagues and communities.

If you're wondering what to do next, my recommendation is simple.

Register for the AI Festival

In a world where most of us barely have time to think between meetings, emails and competing priorities, where else do you get an entire week dedicated to listening, exploring, imagining, challenging assumptions and learning from others who are navigating exactly the same questions?

See you there.