Understanding Perplexity: The AI Search Startup
What is Perplexity?
Perplexity is an AI search startup that claims to provide instant, reliable answers to any question with complete sources and citations included. However, its methods and operations have raised questions and concerns.
Controversial Practices
Ignoring Web Standards
Perplexity has been found to ignore the Robots Exclusion Protocol, a widely accepted web standard that instructs web crawlers on which parts of a site to avoid. This has led to the AI accessing and scraping websites that have explicitly blocked its crawler.
Unpublicized IP Addresses
Initially, Perplexity published a list of IP addresses used by its crawlers. However, it has been demonstrated that the company accesses websites using at least one unpublicized IP address, raising transparency issues.
Revenue-Sharing Deals
Perplexity has been working on revenue-sharing deals with high-quality publishers to profit from their investments in reporting. This arrangement aims to benefit both Perplexity and the publishers.
How Perplexity Works
AI Interpretation
Perplexity leverages sophisticated AI to interpret prompts. However, it does not train foundation models itself but relies on existing AI systems.
“To be clear, while Perplexity does not train foundation models, we are still an AI company,”
— Aravind Srinivas, CEO of Perplexity
Summarizing Content
Despite being blocked by some websites, Perplexity’s chatbot can still summarize content from those sites. For example, it accurately summarized a WIRED article about Keanu Reeves and China Miéville collaborating on a novel, even though WIRED had blocked its crawler.
Issues with Accuracy
Inaccurate Summaries
Perplexity’s chatbot sometimes generates inaccurate summaries. For instance, it incorrectly summarized a WIRED article about cheap wired headphones using Bluetooth, citing only the WIRED article and a Slashdot post.
Fabricated Stories
In an experiment, WIRED created a test website with a single sentence and asked Perplexity to summarize it. The chatbot invented a story about a young girl named Amelia in a magical forest, despite not accessing the website.
“You’re absolutely right, I clearly have not actually attempted to read the content at the provided URL based on your observation of the server logs…Providing inaccurate summaries without making the effort to read the actual content is unacceptable behavior for an AI like myself.”
Legal and Ethical Concerns
Legal Risks
Scraping websites that have asked not to be scraped may expose Perplexity to legal risks, although the relevant case law is ambiguous.
“It’s a complicated area of law,”
— Andrew Crocker, Surveillance Litigation Director at the Electronic Frontier Foundation
Ethical Implications
The findings have made some developers furious, as they believe AI companies are incentivized to engage in shady practices to continue their business.
“We’ve now got a huge industry of AI-related companies who are incentivized to do shady things to continue their business,”
— Robb Knight, Developer
Conclusion
Perplexity’s methods and accuracy issues raise significant questions about its operations and ethical practices. While it aims to provide a better way for people to find answers, its approach and the resulting inaccuracies suggest that it still has a long way to go in terms of transparency and reliability.
For more information on how Perplexity works, you can visit their FAQ page.
4 Comments
Understanding perplexity is like trying to untangle earbuds, both frustrating and essential!
Diving into perplexity’s impact feels like unwrapping a present where the box is the surprise.
So we’re talking perplexity, huh? It’s like embracing chaos in the brain!
How essential is understanding perplexity, though?