Have you had customer blowback when they found out you were using AI?

[AbrahamJaffar]

I recently talked to a chatbot which was pretending to be a real person. I can smell AI responses a mile away and when I confronted it. It doubled down. This instantly made me distrust the person. As AI adoption increases there will be blowback.

How are you handing this at your dealership? Has this occurred recently?

 

[MakaiMorais]

I recently talked to a chatbot which was pretending to be a real person. I can smell AI responses a mile away and when I confronted it. It doubled down. This instantly made me distrust the person. As AI adoption increases there will be blowback.

How are you handing this at your dealership? Has this occurred recently?

View attachment 11533

I'm on the vendor side, I build these for stores rather than run one, so answering from that angle. We have ours disclose. It says it's an AI assistant if anyone asks and we stopped letting it dodge the question. The doubling down is what burns trust, not the AI itself. Most people are fine texting a bot to see if a car's still on the lot, they're not fine finding out they were lied to about who they were talking to. The other thing worth deciding early is what it does when it can't answer, because a bot that improvises instead of handing off to a human is where the bad screenshots come from.

 

[carguyfiles]

We took the opposite route from most vendors in this space: our tool doesn't talk to buyers at all. It posts the cars, keeps the prices right and pulls sold units, and a human answers Messenger. Partly because the reply is where the deal is and I'm not handing that to a model, partly because of exactly what Abraham ran into. A bot that doubles down about being a person is a trust fire you don't recover from.

Where we do use AI is photos, and the rule there is the same one Makai is describing for chat: it can't lie. The background can change, the car can't. A model that repaints a silver car a shade lighter or drops a badge has just misrepresented the vehicle, and the buyer finds out on the lot. Disclose the bot, constrain the model, keep a human on anything that touches money or the truth about the car. That's the whole policy.

 

[KennySmithAutoUnite]

Makai and Joe are hitting the part that matters most to me. From the dealer side, the customer doesn’t care which vendor built the bot. If it lies, gives bad information, or pretends to be a person, my store owns that experience.

My line would be pretty simple. AI can help the customer, but it can’t fake being one of my people. If the customer asks what it is, answer clearly. No dodging it. No fake identity. No doubling down.

I also think the guardrails have to go beyond just telling someone it’s AI. Research, comparisons, features, basic questions, helping somebody narrow down what they need, those are all good uses for it.

But when it gets into a specific vehicle, price, availability, incentives, payments, trade value, financing, or anything else that can cost the customer or the dealership money, it either needs verified information or a person needs to step in.

I would much rather have a bot say, “I don’t know, let me get someone,” than confidently make something up. One bad answer can undo a lot of trust very quickly.

I don’t think dealers should run from AI. Customers are already using it every day. We just need to be smart about where we use it and what authority we give it.

That’s how I look at it. Use AI to make the customer’s experience easier, but keep the accountability with the store.