zlacker

[parent] [thread] 2 comments
1. tdulli+(OP)[view] [source] 2025-07-07 13:18:27
Author here. I am entirely ok with using "goal" in the context of an RL algorithm. If you read my article carefully, you'll find that I object to the use of "goal" in the context of LLMs.
replies(1): >>Timwi+Aq2
2. Timwi+Aq2[view] [source] 2025-07-08 12:43:55
>>tdulli+(OP)
If you read the literature on AI safety carefully (which uses the word “goal”), you'll find they're not talking about LLMs either.
replies(1): >>tdulli+oF2
◧◩
3. tdulli+oF2[view] [source] [discussion] 2025-07-08 14:30:29
>>Timwi+Aq2
I think the Anthropic "omg blackmail" article clearly talks about both LLMs and their "goals".
[go to top]