7 Comments

Is there an easy example of a non-conscious entity with morally relevant agency?

Expand full comment
author

not an existing one! which is why this is such murky territory; these things seem to go together in biology but could come apart in AI

https://experiencemachines.substack.com/p/ai-systems-as-real-life-thought-experiments

Expand full comment

You might find Nicolas Delon's work on this front interesting - he suggests v.early / v.simple animals might be examples. I had the pleasure of talking to him here https://youtu.be/f6hM34jEHzU?si=DlOTrL47PcuM9ffE (The "Who Matters?" section 29 mins in is most relevant). Also on audio Sentientism podcast if you prefer listening. His papers are fascinating if you want to delve deeper too: https://nicolasdelon.com/

Expand full comment
author

Thanks so much Jamie! I'm a big fan of Delon's work and have met him around in this space. But I do need to delve deeper and welcome the pointer / reminder!

Expand full comment

Aw shucks

Expand full comment

There is solid evidence (IMO) of unconscious agency is humans, and presumably in other animals as well (it would be odd for agency to be systematically conscious in nonhumans but not humans). This doesn’t mean there are creatures who have agency but are not conscious but it suggests a possible route and (again IMO) makes it plausible there may be nonsentient agents either in nature or artificially created.

Expand full comment

Well I don't agree with how those terms were defined, but I do think anyone trying to build machines that can act in a manner even slightly outside of "dumb" automation should REALLY consider the implications. Given that most of ML/DL as an industry, not to mention every investor, is foaming at the mouth over millions or even billions of "installs" or users, I'd say they have already failed.

Expand full comment