AI Can Hand You the Answer. It Can't Hand You the Practice.

This week we sat down with Helen Routledge, CEO of Totem Learning, who has spent a career at the join between psychology, behaviour change and serious games — and who has some pointed things to say about what happens to learning when the answer is always one prompt away.

🎧 Short on time? Hit play on the full episode here — it's a proper deep dive on serious games, AI-generated content, and why the doing still matters.

Helen's central point is a simple one that lands harder the longer you sit with it. Knowing a thing and being able to do a thing are not the same, and only one of them can be handed to you. Mental models — the internal maps that let an experienced person walk into a mess and know roughly where to look — get built through repetition, friction and the occasional cock-up. AI is very good at removing exactly those three things. Strip out the struggle and you strip out the learning that was hiding inside it.

Which is where serious games come in, and why she's spent twenty years on them. A well-designed simulation gives people the reps: a safe place to make the decision, get it wrong, feel something about it, and go again. That emotional bit isn't decoration — it's the thing that makes the lesson stick. We get into how AI changes the economics of building that kind of experience, and where the quality falls apart when nobody is steering.

Because that's the other half of her argument. AI can generate a world, a character, a branching scenario in minutes, and the temptation is to let volume stand in for craft. Helen's take is that the human direction becomes more valuable, not less — someone still has to decide what the experience is for, what emotional truth it needs to carry, and where the learner should be allowed to fail. She's blunt about the risk of AI as a control rather than a support: an AI that quietly steers a learner to the right answer has removed the only bit that was worth doing.

We also cover the ethics and regulation question, whether the pace needs slowing, and the generational split in adoption that isn't landing the way most people assume. And she leaves us with the best game recommendation we've had on the show — Zelda: Breath of the Wild, for reasons that turn out to be genuinely about nested problem-solving.

Why it matters: every organisation rolling out AI is implicitly betting that faster answers make better people. Helen's work suggests the opposite unless you design for it — the skill you don't practise is the skill you lose, and the firms that stay sharp will be the ones that deliberately keep the reps in.

👉 Listen to the full conversation with Helen Routledge — it's worth your commute.

Enjoying Project Flux? Subscribe here to get our take on AI and the built environment in your inbox every week — free.

Project Flux is editorially independent. We're not paid to feature any guest, tool or company — we share what we find genuinely useful and let you make your own mind up.