3 ms·

I’m saying that when you start by dismissing a large group of people with a range of views as hysterical it’s pretty clear that you’re not here for any intellec

by acdha 5mo ago

I’m saying that when you start by dismissing a large group of people with a range of views as hysterical it’s pretty clear that you’re not here for any intellect reasons but rather just to defend a technology which you’ve incorporated into your self-identity.

simianwords 5mo ago
“The accuracy issues are real” my main point was that ChatGPT is more accurate than Google or kagi. If you can’t even believe this or you need sources, there’s no debate. You are ignorant of the strengths of LLMs.
- acdha 5mo ago
  That’s not just wrong but a category error[1] and it also misses the other arguments. The fact that you think wanting sources is a problem is a rather perfect demonstration that you aren’t approaching this carefully, though. Ironically, if you ask an LLM this question it gives a much better answer than you have so far, although both of the ones I used still miss the larger impact on the web. 1. A search result’s accuracy is attributed to the source, whereas an LLM’s summary is attributed to the LLM—especially in the cases where it “cites” sources which don’t make the claims it emitted.
  - simianwords 5mo ago
    > 1. A search result’s accuracy is attributed to the source, whereas an LLM’s summary is attributed to the LLM—especially in the cases where it “cites” sources which don’t make the claims it emitted. This makes no sense. Don’t ask the llm for summary. Ask it for the links directly. In what way are they different? ChatGPT is a superset of a search engine. Here’s an example of me using it: https://chatgpt.com/share/6a159c95-3bfc-832a-bfed-52188a548d88 https://chatgpt.com/share/6a159c95-3bfc-832a-bfed-52188a548d... If your fundamental problem is the accuracy of ai overview I’m with you. But AI mode works well.

3 ms·

I’m saying that when you start by dismissing a large group of people with a range of views as hysterical it’s pretty clear that you’re not here for any intellec

by acdha 5mo ago

simianwords 5mo ago
“The accuracy issues are real” my main point was that ChatGPT is more accurate than Google or kagi. If you can’t even believe this or you need sources, there’s no debate. You are ignorant of the strengths of LLMs.
- acdha 5mo ago
  That’s not just wrong but a category error[1] and it also misses the other arguments. The fact that you think wanting sources is a problem is a rather perfect demonstration that you aren’t approaching this carefully, though. Ironically, if you ask an LLM this question it gives a much better answer than you have so far, although both of the ones I used still miss the larger impact on the web. 1. A search result’s accuracy is attributed to the source, whereas an LLM’s summary is attributed to the LLM—especially in the cases where it “cites” sources which don’t make the claims it emitted.
  - simianwords 5mo ago
    > 1. A search result’s accuracy is attributed to the source, whereas an LLM’s summary is attributed to the LLM—especially in the cases where it “cites” sources which don’t make the claims it emitted. This makes no sense. Don’t ask the llm for summary. Ask it for the links directly. In what way are they different? ChatGPT is a superset of a search engine. Here’s an example of me using it: https://chatgpt.com/share/6a159c95-3bfc-832a-bfed-52188a548d88 https://chatgpt.com/share/6a159c95-3bfc-832a-bfed-52188a548d... If your fundamental problem is the accuracy of ai overview I’m with you. But AI mode works well.