Sign in

Rhys Southan

@rhysso.bsky.social
58 followers 102 following 17 posts

Philosophy PhD student at the University of Oxford

PostsRepliesMedia
Rhys Southan @rhysso.bsky.social · 17/09/2025
New Works in Philosophy posted our summary of "A Timing Problem for Instrumental Convergence" newworkinphilosophy.substack.com/p/a-timing-p...
newworkinphilosophy.substack.com
“A Timing Problem for Instrumental Convergence” - Rhys Southan, Helena Ward, and Jen Semler (all @University of Oxford)
Philosophical Studies, 2025
030
Reposted by Rhys Southan
Jakob Lohmar @jakoblohmar.bsky.social · 08/07/2025
My paper on "The Moral Importance of Low-Welfare Species" is now available online first: journals.publishing.umich.edu/jpe/news/207/ It discusses whether benefits for species with small welfare ranges can outweigh significant human goods.
journals.publishing.umich.edu
The Moral Importance of Low-Welfare Species. Forthcoming article by Jakob Lohmar
Abstract: Many species seem to have much smaller welfare ranges than we do. Which importance should we assign to the welfare of these low-welfare species when we have to decide …
161
Rhys Southan @rhysso.bsky.social · 08/07/2025
So really, I think your objection is the goal-first objection, not the delay objection
010
Rhys Southan @rhysso.bsky.social · 08/07/2025
So, there's nothing necessarily irrational or mistaken about having a goal that humans settle Mars and also wanting future people to have that goal. The point is just that there is nothing instrumentally irrational about abandoning that goal, because as soon as you do, it's not your goal anymore
100
Rhys Southan @rhysso.bsky.social · 08/07/2025
I realized I responded to this in a confusing way. What you say here is consistent with the timing problem argument in the paper. The timing problem just entails that if you abandon your goal, you have not done something that violates your goal, since it's not your goal anymore...
100
Rhys Southan @rhysso.bsky.social · 04/07/2025
I think because I have a hedonic-phenomenal notion of desires, and we couldn't assume that a superintelligence will have desires in that sense. In the paper as it is, your objection is closest to the Delay Objection
000
Rhys Southan @rhysso.bsky.social · 04/07/2025
At some point I was thinking about having a potential objection be that we are relying on a very strict concurrentist notion of desire satisfactionism (with the objection being that this is controversial, or at least debatable), but I changed my mind about this
100
Rhys Southan @rhysso.bsky.social · 04/07/2025
So, it's only a problem if you and your future self exist at the same time, which by definition you don't
000
Rhys Southan @rhysso.bsky.social · 04/07/2025
You prefer your future self to have value X concurrently with X being your value, and are indifferent about your future self having value X if X is not your value. If you abandon the value X, your future self will not value X, but this will not go against your value of X, since you no longer value X
100
Rhys Southan @rhysso.bsky.social · 04/07/2025
You prefer others to share your value X while you have the value X, but are indifferent about whether they have the value X when X is not your value. Same with your future self....
200
Rhys Southan @rhysso.bsky.social · 04/07/2025
I had an additional thought after our chat. I wonder if you see wide scope as collapsing into narrow scope, whereas we see narrow scope as collapsing into wide scope
110
Rhys Southan @rhysso.bsky.social · 03/07/2025
Thanks!
000
Rhys Southan @rhysso.bsky.social · 03/07/2025
Will a superintelligence preserve its goals just because that makes it better at achieving its goals? We argue no: link.springer.com/article/10.1...
link.springer.com
A timing problem for instrumental convergence - Philosophical Studies
Those who worry about a superintelligent AI destroying humanity often appeal to the instrumental convergence thesis—the claim that even if we don’t know what a superintelligence’s ultimate goals will ...
141
Rhys Southan @rhysso.bsky.social · 16/04/2025
Has anyone written about whether Barron Trump's being a student at NYU protects NYU against Trump's attacks on universities? Last I checked, NYU wasn't on Trump's list of universities to defund
010
Rhys Southan @rhysso.bsky.social · 17/03/2025
Nice takedown of the phrase "moral blackmail": benthams.substack.com/p/the-phrase...
benthams.substack.com
The Phrase "Moral Blackmail" Has Become A Figleaf For Psychopathy
As has its twin, the phrase emotional blackmail
020
Rhys Southan @rhysso.bsky.social · 15/03/2025
Every time I get a head rush, I think about whether my life would be worth living if I had a head rush every waking moment, and I can never settle on an answer
020
Rhys Southan @rhysso.bsky.social · 19/12/2024
Hehehe good thing no one can see what I'm writing in my secret scratchpad
040
Rhys Southan @rhysso.bsky.social · 19/11/2024
"Jonathan Birch is for Terminator/Hal 9000. Eliezer Yudkowsky is for you." www.theguardian.com/technology/2...
theguardian.com
AI could cause ‘social ruptures’ between people who disagree on its sentience
Leading philosopher says issue is ‘no longer one for sci-fi’ as dawn of AI consciousness is predicted for 2035
030