Henry Shevlin(@dioscuri) 's Twitter Profileg
Henry Shevlin

@dioscuri

Philosopher of AI ethics, cog sci, animal minds, & consciousness. Associate Director @LeverhulmeCFI, University of Cambridge. 🌱 πŸ‡ΊπŸ‡¦πŸ‡¬πŸ‡§πŸ‡ͺπŸ‡ΊπŸ‡΅πŸ‡­

ID:17308990

linkhttp://www.henryshevlin.com calendar_today11-11-2008 12:52:12

5,5K Tweets

4,0K Followers

3,0K Following

Henry Shevlin(@dioscuri) 's Twitter Profile Photo

Which is axiologically preferable: (i) a world of n maximally happy sentient beings all having phenomenologically identical experiences, or (ii) a world with n slightly less happy (but still very happy) sentient beings having phenomenologically varied experiences?

account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

This doesn't happen much in philosophy these days but it's still dismaying to see it when does occur. The vast majority of papers are not written to be read aloud, and most philosophers are indifferent orators.

account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

Feels kind of sus that the biggest animal ever (whales), most technologically advanced animal ever (humans), and fastest animal ever (peregrine falcons) happen to be alive at the same time. I think we should hit pause on evolution while we figure out what's going on.

account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

I was talking (out loud) to ChatGPT today and got onto the topic of AI optimism/pessimism. I tell it I wrote a piece about this a year ago. Cue the clicking noise that means it's searching the web, and it says 'I just read your piece, very interesting, I have some questions...'

account_circle
Sven Nyholm(@SvenNyholm) 's Twitter Profile Photo

Out now: Anna's AI Anthology (edited by Anna Strasser DenkWerkstatt Berlin), featuring my 'Generative AI's Gappiness', along w/ essays by the great Dan Dennett,Keith Frankish, Eric Schwitzgebel, Henry Shevlin, among others. Also, a graphic novel(!) as a bonus feature:
xenomoi.de/philosophie/mo…

Out now: Anna's AI Anthology (edited by @AnnaStrasser1), featuring my 'Generative AI's Gappiness', along w/ essays by the great Dan Dennett,@keithfrankish, @eschwitz, @dioscuri, among others. Also, a graphic novel(!) as a bonus feature: xenomoi.de/philosophie/mo… #aiethics
account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

Sorting through and labelling my old collection of mp3s from the late 90s/early 00s and so far I have found four different spellings of 'Lynyrd Skynyrd'

account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

AI risk hot take: outer alignment failure looks less likely given how good Transformer models + RLHF are at context. Inner alignment failure as worrying as ever. Suggestion: more entry-level fiction showing what that looks like (ideally w/o using concepts like 'mesaoptimizer')

account_circle
Henry Shevlin(@dioscuri) 's Twitter Profile Photo

I endorse a restricted version of the Efficient Market Hypothesis specifically for parking: if the main parking lots are full and you see an empty space, there's usually a catch (disabled only, 30 min max, etc.). My wife is more optimistic, and she's been right more than once.

account_circle