Tim Duffy(@timfduffy):So, we have a couple reasons to expect to be getting emotional representation related to the assistant response, even under the hypothesis that the pain axis works roughly like Anthropic's emotion concepts do: that is, not picking out a ‘self’ representation but rather a ‘current speaker’ representation. And I think this is plausible, given the similar experimental design.

Tim Duffy

@timfduffy

I like utilitarianism, consciousness, AI, EA, space, kindness, liberalism, progressive rock, economics, most people. Substack:

加入 August 2008

819 正在關注 1.4K 粉絲

Tim Duffy@timfduffy

2026.09.24 03:03

So, we have a couple reasons to expect to be getting emotional representation related to the assistant response, even under the hypothesis that the pain axis works roughly like Anthropic's emotion concepts do: that is, not picking out a ‘self’ representation but rather a ‘current speaker’ representation. And I think this is plausible, given the similar experimental design.