Using AI to Shop for a Car? Can You Trust the Recommendations? Ep. 137

Are LLMs like ChatGPT directing car buyers to paid sponsors? On this episode of The AutoGuide Show, we're going inside the hidden mechanics of AI car shopping.

When it comes to buying a new or used vehicle, millions of consumers have shifted from Googling "what car should I buy?" to asking their favorite Large Language Model (LLM). On the surface, asking ChatGPT or Claude for car recommendations feels like having an unbiased, hyper-knowledgeable expert in your pocket—a welcome relief from Google’s search engine results, which devolved long ago into a wall of sponsored links and targeted ads.

But is AI already quietly following the exact same path?

In our latest video podcast episode, we sat down with Corey Lydstone, Founder & CEO of CarGenius AI (and former co-founder of Autolist), to discuss a fascinating discovery. By analyzing tens of thousands of automotive search queries across the U.S. every single day, his team uncovered subtle shifts in how ChatGPT handles car shopping: shifts that raise critical questions for buyers, dealers, and the future of the web.

From Conversational Helper to Structured Sales Funnel

Typically, asking an LLM for advice feels like a dynamic back-and-forth conversation. However, during recent search analysis, Corey’s team noticed ChatGPT behaving differently when users asked for car recommendations.

Instead of generating simple text responses, ChatGPT began explicitly interviewing users through interactive UI elements: asking follow-up questions about price range, features, and crucially, payment preference (cash vs. financing).

When users selected financing options, the user experience shifted again:

• Redirected Traffic: When recommending local inventory, instead of linking directly to local dealership vehicle pages—which often carry free Carfax reports and transparent details—ChatGPT repeatedly routed buyers toward Capital One.
• Coincidental Ads: Directly below these results sat targeted promotional blocks for Capital One.
• Structured UI Elements: Rather than organic, dynamically predicted text, the output featured highly structured design elements that looked designed by product managers rather than pure machine learning prediction.

While LLMs have publicly established data partnerships with platforms like Yelp or Indeed, these guided automotive pathways raise an important question: Are users receiving purely objective advice, or are subtle lead-generation structures already steering consumer decisions behind the scenes?

The Dealer vs. Consumer Dilemma

Corey notes that the traditional automotive marketplace model has long suffered from a structural conflict of interest: platforms cater products to consumers, but get paid by charging dealerships. This dynamic often forces marketplaces to withhold key details from consumers (like free vehicle history reports) to keep traffic locked within their own monetization loops.

Yet, as AI becomes the new "front door" for consumer decision-making, it holds the potential to fix these inefficiencies. Buyers can quickly compare trade-in values, evaluate financing options, and analyze dealership reputations without sifting through dozens of sponsored web pages. And, in the near future, personal AI agents will handle the tedious logistics—scheduling service appointments, querying local inventory for price drops, and negotiating with dealership systems programmatically.

For more on Car Genius AI check out the dealership rankings we discuss on the show here.