The Flaw in Using AI to “Do AI”

AI is powerful. It can write reports, analyse data, generate ideas, and even help you plan a project. Which means it’s tempting - dangerously tempting - to think: why not use AI to figure out how we should use AI?

On the surface, it sounds efficient. After all, who better to explain AI than the technology itself?

But there is a big flaw in that thinking and one that I fell foul of when I started using AI extensively - using AI to “do AI” skips the human thinking that makes AI valuable in the first place.

The Seduction of the Shortcut

A common pattern we see in organisations is this:

1. Leaders want to “get into AI.”

2. They ask an AI tool to draft a strategy, write an adoption plan, craft an email or list “top use cases" - I speak from experience on this!

3. They present the output as progress.

The problem? What comes back is generic, context-free, and often just a rehash of what’s already on the internet. It might look polished, but it’s not tailored to your business, your data, your people, or your culture. It misses the mark when you delve into the output a bit more deeply.

It’s the equivalent of asking a satnav to plan a journey without telling it where you’re starting from.

Why It Doesn’t Work

It's important to note that AI is pattern recognition at scale. It spots trends in data and reflects them back. That’s invaluable when applied to customer behaviour, market signals, or operational processes.

But when it comes to deciding why you need AI, where it fits, and how it should be governed, you need more than a pattern. You need judgement, strategy, and context.

Without that, the risk is:

• AI-wash strategies - generic plans that look good but don’t drive outcomes.

• Misaligned investments - chasing tools and pilots that don’t link to business goals.

• Cultural backlash - teams who see AI as imposed rather than enabling.

• Tactics without a strategy - and tactics without strategy is the quick way to defeat

In short: AI can accelerate the how, but only humans can define the why. Something that I explored in detail before I set Talisman AI up.

The Right Role for AI in AI Adoption

Here’s where AI is useful in building your AI journey:

• Idea generation: surfacing possibilities you might not have considered.

• Scenario testing: modelling different outcomes and options.

• Scaling communication: translating strategy into clear, audience-specific messages.

But those activities should sit within a human-led framework. You still need to set vision, align outcomes, and decide the ethical, cultural, and commercial boundaries. That’s the work AI can’t do for you.

Bottom Line

AI won’t design your AI strategy. It can only amplify the clarity - or add to the confusion - that’s already there.

So before you ask an AI tool to “tell you how to do AI,” pause. Ask yourself: what outcomes do we want, what problems are we solving, and how will this fit with our people and culture?

Get that right, and AI becomes a force multiplier. Get it wrong, and you’re just automating (and accentuating) the flaws.

At Talisman, we help organisations move past shortcuts and build AI strategies that are human-led, outcome-driven, and ethically grounded.

Because doing AI properly starts with people - not prompts.