Clojurians Log v2
ai-skeptics 2026-10-21
Some mistakes are wild. "Stochastic parrot" immediately comes to mind.
Just now, Haiku 5.5 read a file at /home/p-himik/Downloads/xxx.json correctly. Claude put it in the context, however that works.
It then grepped it a couple of times for different things - all worked.
And the third time it tried to grep /home/p-humik/Downloads/xxx.json. Note the "typo".
There was something nagging me in the back of my head for a while and I just couldn't pinpoint it until last night. In this great piece https://clojurians.slack.com/archives/C0B7Q4XRM1U/p1792135400819699 There was this > because I’ve seen how individuals start to lose their ability to use common sense checks on what they’re doing when they get in the habit of trusting AI output. Eureka. This has been bugging me, smart people falling into some kind of rabbit hole and just accepting everything AI outputs. Have you experienced anything similar, like this kind of behavior makes dialogue almost impossible and causes active harm in the form of slop grenade PR's and so forth?
yes, humans are lazy by default (using system2 requires much more energy and effort than just going by gut)... so once you see many correct solutions from the LLM, it is easy to fall into this trap
Yup, my anecdotal experience seems to correlate this with narrow-minded "AI will change the future of humanity" type of drinking the AI hype kool-aid
It feels like belief
to make matters worse, there have been some high profile people online who have made the "I trust the AI" a part of their identity and flaming people who don't
but, perhaps that's just a reflection on typical online spaces and personalities... not necessarily about AI
Same kind of behavioral patterns seem emerge around all hot topics, that is true, so this may be more like a feature of humanity 😅