Why You Shouldn’t Trust AI for Product Advice, Warns Tech CEO
▼ Summary
– Ziff Davis CEO Vivek Shah warns that AI chatbots increasingly cite marketing sources instead of journalism when providing buying advice.
– Shah urges consumers to verify AI chatbot sources, as brands may favor their own products rather than provide neutral information.
– Testing revealed mixed results among popular chatbots, with some showing more vendor sources and others more publisher sources.
– Despite suing OpenAI over intellectual property concerns, Shah remains optimistic about AI’s transformative potential for business.
– Shah suggests his company’s future may include licensing trustworthy data to AI chatbots to ensure reliable information.
When researching significant purchases, many consumers now turn to artificial intelligence for quick answers. However, a prominent technology leader cautions that these AI systems often prioritize marketing materials over impartial journalism, potentially leading shoppers toward biased recommendations. Vivek Shah, CEO of Ziff Davis, emphasizes that while AI offers convenience, users must critically examine the sources behind chatbot responses to ensure they’re receiving balanced information rather than promotional content.
During a recent podcast interview, Shah observed a troubling shift in how large language models gather information. “Ultimately, where we obtain our information matters significantly,” he explained. “If you examine citations in LLM chatbots closely, you’ll notice sources have transitioned from journalistic outlets to marketing materials.” This change raises concerns about whether AI-generated shopping advice serves consumer interests or merely promotes specific brands.
Shah’s warning extends beyond ChatGPT to include other popular platforms like Google Gemini, Perplexity, and Anthropic’s Claude. Although these tools typically cite their information sources, they sometimes make these references difficult to access or verify. The executive encourages users to investigate these citations thoroughly, noting that brand-generated content often favors specific products rather than providing objective analysis.
“I find it remarkable how many citations originate from brands rather than publishers,” Shah commented. “For most companies, being featured in an AI response represents a positive outcome since it promotes their products. However, this arrangement may not align with what’s truly best for consumers. I strongly advise people to examine what’s actually informing the answer and consider whether they would have trusted those sources before AI chatbots existed.”
A practical test involving four major chatbots asking about Meta Ray-Ban Display glasses revealed varying approaches to sourcing. Claude and Gemini demonstrated greater reliance on vendor materials, while Perplexity and ChatGPT incorporated more publisher content. Perplexity stood out by displaying sources most prominently and making them easily accessible. It’s worth remembering that large language models produce different results for each user and can even generate varying responses to identical questions.
Shah’s perspective carries particular weight given his position leading companies that provide vendor-neutral purchasing guidance, including ZDNET. His company’s lawsuit against OpenAI regarding content scraping might suggest an adversarial stance toward AI, but Shah presents a more balanced view. “I’m actually quite optimistic about AI’s potential within our business context, and we’re currently observing some genuinely intelligent implementations,” he clarified.
The CEO acknowledges that licensing reliable data to future AI systems could become part of his company’s strategy. “My concern revolves around intellectual property rights, not a dismissal of AI’s transformative impact on our lives and businesses,” Shah stated. “I firmly believe this technology will prove revolutionary.” This nuanced position recognizes both the challenges and opportunities presented by artificial intelligence, advocating for responsible development while maintaining enthusiasm for its potential benefits.
(Source: ZDNET)
