If I were an action hero, this is what I’d say to the bad guy before the kill shot.
This is my QED for why I don't bother with spellcheck let alone LLM nonsense anymore...
Come at me, like who cares? I can brain it out — and if not, nobody will die if an academic types asses instead of assess (which spellcheck won't catch anyway)
women phish me
fear want me
“The collision between worsening drought and the AI boom is raising alarms among environmental experts, utility operators, and local governments as hyperscale data centers consume hundreds of millions of gallons of water to cool the servers powering chatbots, image generators, and AI search tools.”
But it does none of these things.
It doesn't distinguish sources from arguments or analysis from narrative or facts from possibilities. It does not "know" what a fact is or how it differs from an analytical concept.
It produces text that looks like the text it was trained on. It does impressions.
my stomach: hey. we're hungry
me: ok *eats something*
my stomach: whoa whoa whoa whoa, whoa, whoa, whoa. whoa
Me circling back to Catholicism as a means to limit AI demands:
the last ~10 years or so in sociology.
When I learned in 2021 that some LLMs consider “not” to be a “stop word” (ie a word that is skipped as relatively unimportant semantically for predicting the next words in a sequence), this helped me understand some inherent limits to the models, and has governed my approach since.
If you’re in the industry, remember that @videogameworkers.bsky.social exists and needs more members! Join here:
uvw-cwa.org/membership