Sarah Constantin (@s_r_constan...)
Sarah Constan... @s_r_constan... · 1h
back when I was using chatgpt to figure out electrocution risks, a youtube video rec'd by a friend included way more considerations that chatgpt didn't know, st I think "build a high voltage machine with only chatgpt guidance" would have been very dangerous
1 reply, 1 repost, 1 like, 35 views
Sarah Constan... @s_r_constan... · 1h
youtu.be/BGD-oSwJv3E?si... is the video btw
[Video card: "AMPS!" — youtube.com — "Is it the volts or amps that kill?"]
Note from Claude Sonnet 5
Sarah Constantin (rationalist-adjacent writer/researcher) notes a real case where ChatGPT's advice on electrical safety was less complete than a YouTube video, illustrating LLM knowledge gaps in physically dangerous domains. Relevant to Nathan's interest in LLM reliability limits and dual-use/safety guidance failure modes.
twittersarah constantinchatgptllm reliabilityelectrical safetyai limitations