|
|
|
|
|
|
by tadfisher
28 days ago
|
|
|
That there's a possibility of any of this doesn't make it true. The same set of weights might cause the output here to look like the tool is "caring", and at the same time some output for another prompt might cause the output to be a recommendation to exterminate humans starting with the most power-hungry civilizations. You are failing to be a reasonable judge for the Turing test. The LLM is not capable of "caring". |
|
|