AI startup aims to disrupt search business; Forbes claims they’re ripping them off

The artificial intelligence startup Perplexity AI has secured substantial funding from notable tech investors, including Jeff Bezos, aiming to challenge Google’s dominance in the search engine market. However, Perplexity’s AI-driven search chatbot is already encountering hurdles as news media companies object to its business practices, and tech giants like Google and Apple integrate similar AI features into their products.

Perplexity CEO Aravind Srinivas has been defending the company following an incident where a summarized news story was published with information and similar wording to a Forbes investigative article, without proper citation or permission. Forbes later discovered similar cases with other publications. Additionally, the Associated Press found another Perplexity feature fabricating quotes from real people, including a false quote attributed to a former town official from Martha’s Vineyard.

Srinivas emphasized that Perplexity aims to build positive relationships with news publishers to ensure their content reaches more people. He denied that the company copies content, stating that they aggregate information generated by other companies’ AI systems and provide proper attribution. However, he acknowledged Forbes’ feedback and made changes to highlight sources more prominently.

Perplexity is also exploring revenue-sharing partnerships to compensate news publishers when their content is referenced. Randall Lane, chief content officer of Forbes Media, described the dispute as a significant moment in the discussion about AI and stressed the importance of respecting the work of journalists. Lane criticized Perplexity for acting like a media company without proper acknowledgment of original reporting.

Srinivas, a former AI researcher at OpenAI and Google, co-founded Perplexity in 2022. Inspired by his love for Wikipedia, he envisioned Perplexity as a blend of Wikipedia and ChatGPT, providing clean, sourced answers to user questions without the clutter of traditional search results.

Perplexity offers a subscription for premium features and plans to introduce an advertising-based service. While not yet profitable, the company claims to be more sustainably run than others because it does not train its own AI models, instead relying on existing models from OpenAI, Anthropic, and Meta Platforms.

One Perplexity feature, Writing, generates text without web searches, but it has produced unsourced and fabricated content, highlighting the challenges of AI “hallucinations.” Srinivas noted that this feature is a minor use case and more prone to errors since it doesn’t use the web search capabilities of the main product.

The debate around Perplexity underscores the broader challenges for online content creators and journalists, as aggregators depend on the existence of original publications. Stephen Lind, a professor at USC’s Marshall School of Business, pointed out that many companies are rolling out AI services without fully considering the implications or best practices.

Lind acknowledged that Perplexity’s efforts to address feedback are positive but suggested that such considerations should have been integrated from the start. The situation exemplifies the complex dynamics of using AI in content creation and the need for thoughtful implementation and respect for original work.