2026-10-14 12:31 UTC
@ADHDruid@infosec.exchange And again with that fucking anthropomorphised language — ‘chain-of-thought’ (there is no thought), ‘reasoning’ (there is no reason involved), ‘honesty’ (implying that the LLM can lie if it ‘chooses’ to).
I guess that when these people receive email newsletters with personalised greetings (“Hello, Alex! Thanks for reading!”) they’re going to think the email server ‘knows’ them and ‘is aware’.
🙄
Replies (1)
-
@tusharhero@mathstodon.xyz 2026-10-14 18:32
@morrick@appdot.net @ADHDruid@infosec.exchange funny that it "knows" after we have enough trying data with llm chain of thoughts