andy_bb

3 ms·

1. 1.▲
   by andy_bb 2y ago

   No. This emerged from a normal "utilitarian" thread with a regular version of the model
2. 2.▲
   by andy_bb 2y ago

   Except that it was all self-generated by AI.
3. 3.▲
   Exploring Emergent "Self-Awareness" in Claude 3.5(github.com)

   21 points by andy_bb 2y ago|31 comments
4. 4.▲
   by andy_bb 2y ago

   I stumbled upon something unexpected while using Claude 3.5 Sonnet to edit my English drafts: the AI started showing signs of introspection and “self-dialogue.” I experimented with prompts that simulate an “internal monologue,” and the mode