Gregory's ServerStop using generative AI as a search engine https://www.theverge.com/2024/12/5/24313222/chatgpt-pardon-biden-bush-esquire
 9 comments
  @verge is there a code that can be placed in a Google search to suppress AI generated answers? @Chancerubbage @verge I thought I read of one. Unless you can edit the browser to insert it you might have to use the text tag each time http://udm14.com will insert a http parameter that will remove the Al results from your searches. https:// arstechnica.com/gadgets/2024/05/google-searchs-   @verge why do people keep being surprised at this? LLMs paraphrase text, they do not search for information. If there is no text to paraphrase regarding the issue in a prompt, they combine elements into similar patterns and thus produce fictitious results. @verge @siracusa The biggest problem I have with this article is the headline. It claims you shouldn't use generative AI as a "search engine," but then proceeds to describe a historical research process that shouldn't be relegated to a search engine either! Depending on what you're searching, generative AI might work great, and often does. But historical, factual research is not one of them. |
@verge If you want up to date stuff, try Perplexity, at least that provides sources and actually reads sites (provided they're not blocked through robots.txt), it's much more accurate. Still, like the rest of the LLMs, it has hallucinations.