Y
Hacker News
new
|
ask
|
show
|
jobs
OpenAI model modifies shutdown script in apparent sabotage effort
(
theregister.com
)
5 points
by
sausajez
383 days ago
1 comments
abetaha
383 days ago
That is interesting, and reminds me of self preservation. I wonder what training sets elicit that behavior, perhaps most likely all the sci-fi novels.
link