Meet on the ground floor of QNC, on or near the gray slab benches at 7:20ish, and a regular will let everyone into a room in IQC for 7:30 pm. You can see the precise location on plus.codes. The benches are near the Ring Road entrance, and across the hall from an orange-tinted lab space. As always, vegetarian snacks will be provided.
Activity
This week we’ll be reading a selection of posts from the Sequences: Eliezer Yudkowsky’s massive collection of blog posts on rationality from early in the movement’s history, which generally serves as the foundation of the community.
Some of you may have read the Sequences before, but many of you likely haven’t, and for those of us who have, it has in many cases been quite a while. So, we’re doing a refresher! This week, we’re reading Fake Preferences, which is about failed attempts at theores of value.
Readings
The sub-sequence can be found here, but if you’d like individual links to the readings, they are:
The readings are relatively short, and it should only take about a half-hour to read through all of them, so don’t be intimidated by how many there are.
Discussion questions
Why is a clean one-line value ("I just want to be happy," "I'm purely selfish") so much more attractive than the messy truth? Is the pull social, cognitive, or something else?
People tend to adopt moral or value theories that conveniently never require them to change anything. Can you name a value you hold that actually costs you something, and how do you know it's not also a performance?
Everyone in the room is a human, so every intuition about "what minds want" comes from one data point. Where have you been most surprised by another person's values, and what does that imply about non-human minds?
A lot of the posts argue by dialogue with an imagined opponent who always folds. Is that a legitimate way to establish a philosophical point? Would you steelman the opponent differently?
The sequence frames all of this as a prerequisite for building AI that does what we want. Is the material still important for ordinary people, sans the AI motivation? And if so, in what way?
Meet on the ground floor of QNC, on or near the gray slab benches at 7:20ish, and a regular will let everyone into a room in IQC for 7:30 pm. You can see the precise location on plus.codes. The benches are near the Ring Road entrance, and across the hall from an orange-tinted lab space.
As always, vegetarian snacks will be provided.
Activity
This week we’ll be reading a selection of posts from the Sequences: Eliezer Yudkowsky’s massive collection of blog posts on rationality from early in the movement’s history, which generally serves as the foundation of the community.
Some of you may have read the Sequences before, but many of you likely haven’t, and for those of us who have, it has in many cases been quite a while. So, we’re doing a refresher!
This week, we’re reading Fake Preferences, which is about failed attempts at theores of value.
Readings
The sub-sequence can be found here, but if you’d like individual links to the readings, they are:
The readings are relatively short, and it should only take about a half-hour to read through all of them, so don’t be intimidated by how many there are.
Discussion questions
Posted on: