tbabb

generally this pattern makes me update towards "yes", regardless of how I was feeling before. this feels like a stupid heuristic, but it's quite potent

anyway, AI discourse is like this

ballot arguments in favor of no, by Jim Fartswain, unemployed: YOU SHOULD VOTE NO BECAUSE THIS IS PART OF THE FASCIST DEMOCRATS PLAN TO CONTROL EVERY ASPECT OF OUR LIVES AND PUT CHIPS IN OUR BRAINS

ballot arguments in favor of yes, by Paul J. Steinberg, chief expert at Serious Respectable Institution: [long, measured argument in well-written english, including discussion tradeoffs and counterpoints]

[endorsed by 11 organizations, including some recognizable trustworthy / prosocial ones]

starting to think that a large fraction of people want to live in some kind of big pretend game, like children

ok, nevermind, after seeing the full res video, I think it's real

browser tab color serendipity

and the spilled soup. just appears as a tidy spot in a few frames. no physics at all

oh wait. the economy is going to crash because of trump's idiot policies, and they're going to blame AI

What stage is this?

agree. I'd use this to describe their tendencies and (emergent) motivations, but they are still a bit too weak on planning to be placed directly alongside humans at the same stage

I think LLMs are probably around Kohlberg stage 3 right now

I agree that LLMs do not (yet) have enough agency to really be good or evil; they're too naive and suggestible. AI children, of sorts

I don't think that will be the case forever

my categories are moral, amoral, immoral

amoral are things that happen without regard for what is good and evil. and I kind of think humans are default-mode amoral

distinct from immoral, wherein a positive choice to eschew empathy is made

wishing you well. adulthood can sometimes bring new perspective on who our parents are

oh no, I like small models. the big ones scare me

- violin cat
- early controlNet images (e.g. @deepfates' town spiral)

unsure of number 3, maybe traumatized mr. incredible or some other subtle but pervasive AI generation

finally got around to peeling off the chinese numerical symbol for "death" from the car which had a catastrophic engine failure

at some point I realized it's probably not a coincidence that truth is such a core value for me, and I have a parent who lies all the time

the timeline avian bromance fans have been anticipating since the season began

"if someone cuckolds their spouse, we want to be sure that it is very difficult for them to be discovered"

bizarre set of values this would seem to be protecting

"50% credence that UFOs are literally aliens" puts him fully inside my ignore bucket

elaborate?

side point, but there are a few accounts out there that confuse me (emmett's is one) because they have five or six figures of followers but their posting style / engagement is more like they're an order of magnitude less

every week I planted one circle, at the end of my career I had this,

you and me, if we coöperate

We are so far away from peak software

mfw it was me in charge all along

I think this particular discourse has been a case of the following