Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Anthony Accomazzo @ acco.io | URLs make ChatGPT stupid

Feb 27, 2024 - acco.io
The author discusses an issue with ChatGPT (v4) where it fails to access and extract data from a specific webpage as requested. The author suspects that this is due to the AI system caching its response to the page rather than the page itself, as a cost-saving measure. The AI system also seems to ignore requests to perform operations on a URL, providing generic summaries instead of the specific information asked for.

The author tests this theory by saving the webpage as a PDF and uploading it to ChatGPT, which successfully processes the request. This suggests that the AI system is more willing to access and analyze files directly uploaded to it, rather than accessing them via a URL.

Key takeaways:

  • ChatGPT (v4) did not access the webpage when asked to summarize specific content, instead providing a generic summary.
  • The author theorizes that this may be due to a cache system, where the AI avoids web browsing due to its expense and instead caches its response to the page.
  • ChatGPT did not automatically summarize a URL when dropped in, but seemed to ignore requests to perform operations on the URL.
  • When the author saved the webpage as a PDF and uploaded it to ChatGPT, the AI was able to interact with the content as expected.
View Full Article

Comments (0)

Be the first to comment!