should you? · level 1.

the anxiety is rational.

If you're worried about AI and your job, I want to say something clearly before anything else: that worry is rational. It is not catastrophising. It is not a failure of adaptability. It is not evidence that you're "not a digital person."

It is a reasonable response to incomplete information about a genuinely significant technological shift, in an environment where the people who stand to benefit most from your enthusiasm are also the people controlling the narrative.

The reassurance you're being offered — that AI will create more jobs than it destroys, that your expertise is safe, that the tools are just tools — is not necessarily wrong. But it is also not disinterested. The people saying it have reasons to say it that go beyond your wellbeing.

What this module is asking you to do is hold both things at once: the anxiety is rational, and it is not the end of the story. Both are true. The course is about finding out what comes after the anxiety — not by suppressing it, but by understanding it well enough to move through it.

the line everyone's repeating.

"AI won't replace you — someone using AI will."

You've heard this. It's everywhere. And it functions as a remarkably efficient piece of anxiety management — it acknowledges the fear while redirecting it toward individual action. The implication is: if you're worried, the solution is to learn AI faster.

There's something true in it. But there's also something it conveniently doesn't say.

The line assumes that the "someone using AI" doing the replacing is a peer — a colleague, a competitor, a person in the same role who got there first. But the replacement could also be: a smaller team with AI tools doing the work of a larger one. Or a single generalist with AI replacing a specialist. Or a product that didn't exist last year doing what a whole profession used to charge for.

This doesn't mean despair. It means: be honest about what kind of risk you're actually facing, rather than accepting the framing that makes the risk easiest to sell against.

what actually happens.

Most AI implementations fail to deliver what was promised. Not because the technology doesn't work — but because the implementation ignores what was already happening in the organisation, what people actually needed, and what the real barriers were.

The pattern goes like this: a senior team, under competitive pressure and vendor enthusiasm, makes a decision about AI adoption. The decision is announced, often as a strategy or a transformation programme. The people who will be most affected are informed last, if at all. The tools arrive. The training is inadequate. The use cases are vague. The metrics are borrowed from vendor case studies. After 12 months, utilisation is low, nobody can explain why, and the cycle begins again with a new tool.

The organisations that get this right are not the ones with the best technology. They're the ones who asked the honest questions first — including whether AI was the right answer at all — and built from that conversation rather than over it.

shadow ai · the human story.

71% of UK workers have used unapproved AI tools for work. More than half do so weekly. Most are not trying to circumvent security. They are trying to do their jobs.

Shadow AI — the use of AI tools that haven't been officially sanctioned — is one of the most useful data points available to any organisation that wants to understand what's actually happening versus what it thinks is happening.

When people reach for unsanctioned tools, they are telling you something. They are telling you that the sanctioned tools aren't meeting their needs. Or that the conversation about AI has happened above their heads. Or that they've spotted an opportunity their organisation hasn't given them permission to act on.

The response to shadow AI that works is not a policy. It's a conversation. What are people using? Why? What would they stop using if something better were available? What does that tell you about the gap between your AI strategy and your organisation's actual needs?

your expertise is training data.

This is the one people find hardest to sit with.

Every time you use a cloud AI tool for work — every query, every document, every interaction — you are potentially contributing to the training data that will make that tool more capable of doing what you do. The terms of service for most consumer and enterprise AI tools are written to permit this, in language that most people don't read.

This doesn't mean you should stop using AI tools. It means you should know what you're giving when you use them. And it means that for genuinely sensitive professional work — the work that represents your hard-won expertise, your client relationships, your proprietary knowledge — there is a strong argument for keeping that work local.

Local-first AI — running on your own hardware, trained only on what you choose — is not a paranoid response to this. It is a considered one. And it is increasingly accessible.

the question.

The most important question in any project is not "can we?" It is "should we?"

In project management, this distinction is foundational. A project can be technically deliverable, financially viable, and strategically coherent — and still be the wrong thing to do. The "should we?" question is the one that requires the most honesty, involves the most people, and is most often skipped in the rush to begin.

AI adoption has a "should we?" question. It is not: should we use AI at all? For most organisations, in most contexts, the answer to that is: probably, in some form, for some things. The actual question is: should we do this, in this way, at this time, for these reasons, with these people, at this cost?

The Should You? Assessment exists to help you ask this question properly. Not to give you an answer. To make sure the question has been asked, and that the people who need to answer it have been in the room.

what to actually do.

You've stayed with this. That matters. Most people want the comfort before the discomfort — a toolkit, a framework, a set of actions that makes the anxiety feel managed. This course has deliberately not given you that. It has tried to give you something more useful: an honest understanding of what you're actually dealing with.

So what do you actually do?

First: have the conversation you've been avoiding. Whether that's with your team, your leadership, yourself. The questions are now clearer. Use the conversation cards. Start somewhere honest.

Second: audit what's already there. Before you decide what to build or buy, know what's running. The AI Licence Audit tool is free and takes 10 minutes. The Should You? Assessment takes 8. Use them.

Third: consider local-first. For the work that matters most — the work that represents your expertise, your client relationships, your competitive edge — think about whether it belongs in the cloud at all.

This is the end of Level 1. The conversation continues — in the community, in the workshops, in the work itself. Thank you for being honest enough to take it seriously.