In a gripping conversation on CNBC's "Squawk Box", Katie Drummond, Wired's global editorial director, shed light on the magazine's in-depth investigation into Perplexity, an AI search startup. The discussion exposed disturbing aspects of Perplexity's operations, raising crucial questions about AI's impact on journalism and information accuracy.

Perplexity's rapid rise in the tech world, with significant investments from notable figures like Jeff Bezos, has been accompanied by controversy and skepticism. Wired's investigation aimed to demystify Perplexity's operations, search results, and mechanisms.

The findings revealed unclear operational mechanisms, inaccurate responses, and a lack of logical synthesis in its search results. Users reported receiving misleading answers, raising concerns about the validity of Perplexity's search tool.

The investigation also uncovered ethical and legal concerns surrounding Perplexity's data gathering practices. Despite implementing robots.txt protocol, Perplexity allegedly scraped content from publishers without authorization, raising questions about AI companies' interaction with journalistic content and data usage boundaries.

The discussion highlighted the inherent biases in AI responses, emphasizing the "garbage in, garbage out" concept. Perplexity's reliance on accessible web data, which can be biased or inaccurate, means its responses reflect these issues. Drummond argued that direct sources, like reputable journalism outlets, remain the best way to obtain accurate information.

The conversation also explored the disconnect between technology and journalism, with Drummond pointing out that the tech industry has often prioritized advancement over accurate and ethical information dissemination. The investigation raises fundamental challenges in developing AI systems that prioritize high-quality information and highlights the need for a more informed and nuanced approach to AI development.