.png)
Business owners aren't frightened of AI taking over the world. They're frightened of something far more ordinary, and far more likely.
They're frightened that a tool, switched on in their name, will say something they would never have said, to a customer they've spent years looking after.
That's not a sci-fi fear, it's a today fear. And it's worth taking seriously, because in most of the businesses I work with, trust isn't a nice-to-have. It's the whole asset. It's the reason a sceptical buyer chooses you over a cheaper alternative they can't quite bring themselves to risk. Trust is what lowers the perceived risk of buying from you. Lose it, and every other part of your marketing has to work twice as hard.
So the question isn't whether AI is clever enough. It obviously is. The question is what happens when you point something that capable at the most fragile thing you own.
The broom that wouldn't stop
It's an oldie, but if you've ever seen Fantasia you'll remember The Sorcerer's Apprentice. Mickey Mouse is left to fill a well with water, one heavy bucket at a time. It's dull and tiring, so he enchants a broomstick to carry the buckets for him, then falls asleep, pleased with himself. He wakes to a flooded room, a broom that's multiplied into dozens, and no idea how to make any of it stop.
The broom didn't malfunction. That's the part everyone forgets. It did exactly what it was told, faster and more tirelessly than Mickey ever could. The disaster wasn't the broom going wrong. It was Mickey automating a job he hadn't thought through, and then leaving it unattended.
A chatbot on your website is a broom. So is an AI service answering your phone, or a tool firing off replies to customer emails while you're in a meeting. Capable. Tireless. Perfectly happy to keep going without you. The risk was never that it breaks. The risk is that it works, confidently, in your voice, while nobody's watching.
Confident and wrong is the worst combination
Here's the thing about these tools that should give you pause. They don't sound unsure.
My phone has been mispronouncing my surname for years. It says "Cuhn-KEEE" with total confidence and total inaccuracy, and it has never once hesitated. (FYI: It's CON-kee, but people generally get it right first time). That mistake's harmless when it's a name. It's less harmless when the same breezy confidence is applied to a customer's question about what you do.
Because AI would rather be helpful than admit it doesn't know. It's called hallucination, and it isn't a rare glitch, it's a known habit. A customer asks whether you offer a particular service. The bot, eager to please, says yes. You don't offer it. Now there's a customer expecting something you can't deliver, and the relationship has opened with a confident lie told in your name.
And that's the loud failure. The quiet one is worse, because you might never spot it.
An AI handles an enquiry perfectly competently and entirely coldly. No warmth. No reading of the room. None of the instinct that told your receptionist this particular caller sounded worried and needed reassurance before a price. Nothing went wrong, exactly. The customer just came away feeling processed by a company that used to feel personal. They won't complain. They'll simply not come back, and you'll never know why.
That's how trust actually goes. Not in one dramatic incident, but in a slow drift away from the thing that made people choose you. And the more capable the tool sounds, the longer it can drift before anyone notices.
Keep the human at both ends
A CIM webinar I watched recently called it the "AI sandwich".
The idea is simple. You put a layer of human judgment before the AI, and another layer of human judgment after it.
Before, the human sets the brief. What the tool may do, what it must never promise, the moment it should stop and fetch a person. The AI does the fast, repetitive bit in the middle. Then afterwards, a human checks, corrects, and steps in wherever warmth or judgment is needed.
If you've read The Marketing Machine, this will sound familiar, because it's the same principle as every other part of a working marketing system. The machine does the heavy lifting. The human owns the inputs and the quality control. AI is a component in the machine, not the operator of it.
Almost every trust failure I can imagine comes from a missing slice of bread. A chatbot launched with no clear limits. A call service switched on and left alone because it seemed to be coping. The tool didn't fail. It was asked to be the whole sandwich, and no tool can be.
Three questions before you put AI in front of a customer
If you're considering letting an AI tool speak to your customers directly, these three questions will catch most of the trouble before it starts.
What is it allowed to say, and more importantly, what must it never say? If you can't answer that precisely, you're not ready to switch it on.
When something goes beyond its competence, how does a human take over? A tool that hasn't been instructed to hand off gracefully will bluff instead, and bluffing is where trust dies.
Would I be happy for this exact interaction to be the only impression a new customer ever has of us? If the honest answer is no, the tool needs a tighter brief, clearer limits, or a person standing closer to it.
None of this is an argument for staying away from AI. Used well, it takes the repetitive load off your team so they have more time for the conversations that genuinely need a human. The businesses getting this right aren't the cautious ones hiding from the technology. They're the ones who decided, on purpose, where the machine helps and where the human stays, and then kept an eye on it.
Your reputation wasn't built by automation. Don't ask automation to protect it.
If you're working out where AI fits in your marketing without putting that reputation at risk, that's exactly the sort of thing a clarity call is for. Book one with me and we'll work out together where the tool earns its place, and where it doesn't.
Comments