3 ms·
The human mind cannot tell between acquiring new mind-expanding knowledge and acquiring mind-expanding-knowledge-shaped nonsense. We use heuristics to evaluate
by immibis 2y ago
The human mind cannot tell between acquiring new mind-expanding knowledge and acquiring mind-expanding-knowledge-shaped nonsense. We use heuristics to evaluate the quality of input and heuristics can be fooled by adversarial agents. Because the mind can't tell the difference it gives the same pleasure response to both, but the nonsense is much easier to create and wins the race to the bottom.
It reminds me of the way people use LLMs where people can't tell the difference between a good answer and good-answer-shaped nonsense. See this guy's attempt to replicate neetcode.io with an LLM and the neetcode creator's response. The LLM user seems to have fooled themselves into thinking their LLM output is a useful step on the way to recreating neetcode, when it is not: https://www.youtube.com/watch?v=U_cSLPv34xk https://www.youtube.com/watch?v=U_cSLPv34xk