Researcher Pushes Mature AI Safety Debate Past Autocomplete Strawman
Reactions from ranked influencers
2 posts
LLM ≠ AI. A parrot on its own won’t necessarily kill you. A parrot souped up with flamethrowers and flesh-frying laser rifles might. (Even unintentionally.) 🧵
Can we please move past the “fancy autocomplete” strawman of the critiques of AI kills us all and into a more mature discussion of AI safety and security? Most people know by now what mid- and post-training are, and that safety risk means way more than extinction risk. So many takes on Tweeter sadly open with a radical strawman of the critique. You’d hope for more nuance from journalists etc., but not always.