Astonishment: Between Plagiarism Accusations and New Pro Search |

Pro Search is an addition to Perplexity’s Quick Search, which is primarily intended to answer less complex questions quickly and concisely. Pro Search should now be able to filter out irrelevant information. Support for powerful AI models such as GPT-4o (which can also work multimodally) helps the AI ​​Answering Machine with this. For Pro Search – which users without a subscription can use up to five times every four hours – multi-step reasoning was introduced, among other things. This allows users to ask the system to create a plan for answering questions and give instructions in several steps. But other tools also make this possible, such as Copilot and ChatGPT.

In addition, the Perplexity code interpreter has been optimized. Now, code integrations can be processed even faster, allowing users to create content faster or speed up their debugging. With the integration of the Wolfram Alpha engine, even complex mathematical tasks can be solved in no time.

The upgraded Pro Search, which is available to subscribers up to 600 times per day, offers, according to Perplexity,

improved insights from a wider range of trusted sources.

However, some publishers may be bothered by this formulation. Because there is criticism of the content generation of the AI ​​answering machine.

Criticism of web scraping: is content essentially being copied?

Rebecca Bellan asks the question on TechCrunchor Perplexity uses publishers’ content fairly for its own response engine. Wired for example, accuses the companyignoring the quasi-standard protocol for robot exclusion and scraping website content for their own results. Forbes also criticized the AI ​​startupbecause this would have required the content to be copied directly from the publisher for the new pages.

Now, Perplexity doesn’t have to legally follow the Robots Exclusion Protocol — which would greatly diminish its severity, though. The company told TechCrunch that it doesn’t crawl the web per se, but rather summarizes content when, say, a URL is added to the query.

One of Perplexity’s responses to the allegations was to display sources more prominently. This is also clearly visible in the AI ​​search answers; these appear above the text content. At the same time, this only helps publishers to a limited extent; especially if the content is essentially copied. Under US law, Perplexity would probably not be doing anything wrong, but in the EU, the approach would probably be in a gray area. But the company needs content and answers to integrate the planned ads into the context of meaningful and targeted answers for more users. Pro Search is intended to help with this, but the question of, if not legality, then at least reliability of the determined answers and results hangs like a cloud over these updates.

Google competitor Perplexity:
New Discover Feed and first ad

Smartphone against a black background
© Verbijstering, screenshot via Canva

Source link

Leave a Comment