What is Nostr?
Mark M. Evans /
npub1a4v…4tak
2024-11-15 13:38:12
in reply to nevent1q…mptd

Mark M. Evans on Nostr: nprofile1q…ypvmf nprofile1q…80tvd LLMs are over a decade old. They are extremely ...

nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpq5nw5evkeefrzezpznfp28sx7rzyhhdc0gam6rfwhrqjq5qcdrehsaypvmf (nprofile…pvmf) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqgfdrrw9629qxwvwaa3vwnvzhtaqu2c5m9wf6vcwjnmdnhs7hl4jsq80tvd (nprofile…0tvd) LLMs are over a decade old. They are extremely useful but incredibly overhyped. They are the latest silver
bullet for VCs that think being rich makes them smart and eternally optimistic software engineers. Add in CEOs that hate paying people to do work and you’ve got the perfect incentive storm.

LLMs do not “understand“ anything, which means they will always hallucinate. They work great to check people’s work or even to “collaborate” with trained people that understand their field and can push back. But capitalism will flip the model and try make the LLMs replace people and have a small number of people monitor their work. Which humans are terrible at.

The result will be reinforcing all sorts of biases being baked into the data - racism, mysogony, ableism.

A simple example: https://www.geekwire.com/2024/ai-overwhelmingly-prefers-white-and-male-job-candidates-in-new-test-of-resume-screening-bias/
Author Public Key
npub1a4v7n3kwzx7f2p5xh465jgx8fafy9dt6klkwcyt593jzgm04rkkqxq4tak