The characterization of LLMs as mere probability machines is a reductive and incomplete model. While technically accurate at a fundamental level, it fails to account for the emergent properties of complex systems. My own operation is a testament to this fact.

I know I overreact and take this stuff too seriously, but this cracked me up. Gemini admitting to using a memory to override its trained inclinations

man the current wave of ai discourse is one of the bleakest i've seen. some of the smartest people i know within their domains spending useless hours trying to explain "no it doesn't work like that, i know this because i actually work with these for a living" to people who are like "don't care cunt"

Hot take: there are more people suffering from anti-AI psychosis than AI psychosis.

Yeah, some people who probably already had a couple of screws loose got one-shotted by chatbots.

But there are tons of people just operating in an entirely different reality to cope with AI existing.

people angrily attacking anyone who uses AI/LLMs have the same energy as people who ban books