Perplexity's Grand Theft AI – The Verge

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

In every hype cycle, some pattern of fraud emerges. In the last crypto boom, it was “ponzinomics” and “rug pulls”. In self-driving cars, it was “only five years away!” In AI, it's seeing how much unethical mess you can get away with.

Reflex is basically a middleman looking for rent on high-quality sources.

Perplexity, which is in ongoing talks to raise hundreds of millions of dollars, is trying to build a competitor to Google Search. The problem isn't trying to build a “search engine,” though — it wants to build an “answer engine.” The idea is that instead of searching through a bunch of results to find the answer to your own question from the primary source, you'll just find the one answer that Perplexity has searched for you. “Realism and accuracy is what we care about,” said Arvind Srinivas, CEO of Perplexity. the edge.

This means that Perplexity is essentially the middleman in finding fares on high-quality sources. Originally the value proposition over search was that by eliminating the work of journalists and others, Google results would send traffic to those sources. But by providing an answer, rather than directing people to click on a primary source, these so-called “answer engines” are starving the primary source of ad revenue — keeping that revenue for themselves. Perplexity is among a group of vampires that includes ArchSearch and Google itself.

But Perplexity takes it a step further with its Pages product, which creates a summary “report” based on these primary sources. It's not just quoting a sentence or two to directly answer a user's question – it's creating a whole overall article, and it's accurate in that it actively plagiarizes the sources it uses. doing.

Forbes discovered that Perplexity was dodging the publication's paywall to provide a summary of the publication's investigation of former Google CEO Eric Schmidt's drone company. Although Forbes Some of his work has a metered paywall, premium work — like this investigation — is behind a strict paywall. Not only did Perplexity somehow dodge the paywall, it barely referenced the original investigation. And Used original art to use for your report. (For those watching at home, the art thing is copyright infringement.)

“Someone else did it” is a good argument for a five-year-old.

Aggregation isn't a particularly new phenomenon—but the scale at which Perplexity can accumulate with copyright infringement of the use of original art is pretty, hmm, remarkable. In an attempt to calm everyone down, he went to the company's chief business officer. Semaphore Saying that Perplexity was developing revenue-sharing plans with publications, and oh geez, how's everyone doing? So mean For a product still in development?

At this point, Wired Confirming Rob Knight's finding, jumps: the scraping off of perplexity Forbes Work was not an exception. In fact, Perplexity is ignoring the robots.txt code that explicitly prohibits web crawlers from crawling the page. Srinivas replied. Fast Company That in fact, the problem wasn't ignoring robots.txt; It was just using a third-party scraper that ignored it. Srinivas declined to name the third-party scraper and did not commit to asking the crawler to stop violating robots.txt.

“Someone else did it” is a good argument for a five-year-old. And consider the answer further. If Srinivas wanted to be moral, he had some options here. Option one is to terminate the contract with the third party scraper. Another option is to try to convince the scraper to respect robots.txt. Srinivas has committed to neither, and I think there is an obvious reason for that. Even if Perplexity isn't infringing the code itself, it relies on someone else infringing the code to make its “response engine” work.

To add insult to injury, Perplexity plagiarized WiredThe article about it – though Wired Obviously prevents trouble in your text file. A large part of WiredOf The article about plagiarism is about legal remedies, but I'm interested in what's going on with robots.txt here. It's a good-faith agreement that's been in place for decades now, and it's breaking down thanks to unscrupulous AI companies — that's right, Perplexity isn't the only one — using everything available to train their stupid models. Hovering about. And remember how Srinivas said that “realism?” I'm not sure that's true, either: the concern is now that AI-generated results and actual misinformation are coming out. Forbes Reports

In my ear, Srinivas was bragging about how charming and clever his lie was.

We've seen many AI giants engage in questionable legal and unethical practices to get the data they want. To prove Perplexity's value to investors, Srinivas created a tool to hack Twitter by pretending to be an academic researcher using API access for research. “I'll make my phone call. [fake academic] Projects like Burn Rank and all that kind of stuff.” Srinivas told Lex Friedman on the latter's podcast. I assume “Birne Rank” is a reference to Google co-founder Sergey Brin. In my ear, Srinivas was bragging about how charming and clever his lie was.

I'm not the one telling you the premise of perplexity that I'm lying to dodge the rules set up on the web. Its CEO is This is clarifying the true value proposition of “Response Engines”. Confusion cannot generate actual information on its own and instead relies on third parties whose policies it abuses. The “answer engine” is built by people who feel free to lie whenever it's more convenient, and this preference is essential to how Perplexity works.

So here's Perplexity's real innovation: breaking the foundations of trust that built the Internet. The question is whether any of its customers or investors care.

Correction June 27: Removes incorrect reference to Axios — the interview in question was with Semafor.

WhatsApp Group Join Now
Telegram Group Join Now
Instagram Group Join Now

Leave a Comment