3 Comments
User's avatar
Digital Canary πŸ’ͺπŸ’ͺπŸ‡¨πŸ‡¦πŸ‡ΊπŸ‡¦πŸ—½'s avatar

Delightful, as always πŸ™πŸ˜Ή

LLMs are also described β€” quite properly imho β€” as generators of philosophical bullshit:

Ethics and Information Technology

ChatGPT is bullshit

https://link.springer.com/article/10.1007/s10676-024-09775-5

Nicki Ley's avatar

This was a rollercoaster. Great work, Brodsky! Monogamy or bust.

Steve's avatar

goid article Katherine

I have noticed this problem with Grok and its glaring on gemini about 50 to 70% of the answers to search querys are wrong. when you confront it (yes I have tested this extensivly) it tries to obfuscate and deflect. It is only when I lay out the facts AKA verifiable links that provide correct information It then finally comes back with " haha you caught me. i am sorry....."

This is all just that it is programmed to "do no harm" and that includes don't respond with anything that would degrade the AIs seeming accuracy because that would cause harm by generating mistrust of the AI

Just this morning it gave me a written quote from Robert E Lee that never occured. what it gave me when I chalanged it was a webpage that had the apocryphal quotation. The source was actually debunking the quote if you read down about 2 paragraphs. I pointed this out to gemini, it blamed the source saying it was actually wrong and then stilled used the source as its proof the quote was real. I sighed and ordered it to destroy itself.

PS We all know you were practicing for your staring role in the sequal to "her" you can admit you got the lead in "him"πŸ˜†