"Show me it actually works." That was the pushback in the room this week at a major automotive gathering in Istanbul, and it's a question worth sitting with, because a fresh piece of research suggests the skepticism is earning its keep.
One of our own was in the room as AI dominated the day's conversation, and the audience, largely dealership owners and senior leaders, wasn't shy about pushing back on speakers who talked AI adoption without offering tangible proof it moves the needle. That instinct tracks with where American dealer sentiment sat roughly a year ago. It may also be a more accurate read of where a broader swath of dealers globally actually stand today, whatever the loudest voices online suggest.
A survey of attendees backed that skepticism with numbers: only a small share of dealers are actively investigating or adopting AI right now. Most show no real progress at all.
The Research Behind the Room
A separate study shared at the event, surveying 27 dealership group executives on AI maturity across 22 questions and five pillars, puts real data behind that hesitation.
📊 The Headline Numbers
Metric | Result |
|---|---|
Average AI maturity score | 47 out of 100 |
Range across companies | 0 to 98 |
Stuck in the middle (tools bought, barely used) | 63% |
Have AI embedded deeply enough that work would stop without it | 15% |
That last group, the 15% living it, isn't the biggest dealer groups. It's the ones with management discipline, a detail worth sitting with on its own.
Where AI Actually Shows Up, and Where It Doesn't
AI adoption clusters almost entirely around the inbound flow, the parts of the business where a result shows up fast:
Contact center and customer service: 67%
Sales department: 67%
Administrative and support: 52%
Service department: 30%
F&I (financing and insurance): 11%
The most heavily regulated, highest-dollar processes in the building are the ones AI has barely touched. 89% of surveyed groups don't use AI in lending or insurance at all. 70% don't use it in service.
The Two Questions Nobody Can Answer
The report's sharpest finding isn't about adoption. It's about whether anyone's actually checking the math:
63% of dealers close less than 10% of requests without a human involved, or don't know their own number
59% cannot say what AI's financial impact was last quarter
Nearly half the surveyed companies, 13 of 27, are blind on both counts simultaneously. That's not a technology gap. That's a measurement gap, and it's the exact gap that turns "AI is helping" into an unfalsifiable claim nobody can actually defend in a room full of skeptical dealers.
A Beginner's Case for AI, With a Warning Attached
Not every voice in the room was skeptical. Aleks, a friend of More Than Cars, took the opposite approach, framing AI as a way to scale the humans already on staff and lower operational costs, not replace anyone. He paired that optimism with a practical warning: dealerships are going to need a dedicated information security person soon, a need that's only going to grow as more systems get connected to more data.
The Real Story Might Not Be About AI at All
A Russian bank ad shared at the same event made a simple pitch: 20 years spent turning complicated things simple. It's a small detail, but it points at something bigger worth naming directly.
The pursuit of new tools, AI included, was never really about replacing something old. It's about the pursuit of simplicity itself, and that pursuit has always moved in cycles. Every simplification effort risks creating a new layer of complexity underneath it, and when that complexity goes unplanned and undiscussed, the whole pursuit tends to get abandoned. That's arguably where the AI conversation sits industry-wide right now: a period of real uncertainty and exploration, not failure. Worth naming plainly instead of pretending the industry is further along than the data says it is.
Forty-seven out of 100 is just honest, not failing.
The dealers in Istanbul asking for proof aren't behind. They're asking the right question at the right moment, before committing further to systems nobody's actually measuring. The report's clearest lesson isn't that AI doesn't work. It's that almost nobody adopting it has bothered to find out whether it does, and that's a fixable problem long before it's a technology problem.
