@djc more capable at mimicking language without accurately depicting reality with language. better at passing a certain kind of Turing test. that GANs can dream beautifully is an old result. midjourney is remarkable, because what we want from it is dreams and verisimilitude rather than actual accuracy. on the latter, there has been as much disappointment as positive surprise. AI was supposed to have replaced drivers by now.

in reply to @djc

@djc i take much of it at a kind of face value — from my own interactions with that community there has long been a kind of strange sincerity despite the unevaluability of the kind of risks they proclaim. and a lot of it is selection: people who think AI will be powerful select into the industry. but those seeds have now coalesced with industry self-interest in a way that should make the rest of us even more skeptical of their views.

in reply to @djc

@djc i’m sure lots of people are sincere. but i also think it’s not a coincidence that the catastrophists and the communities that now operate big, centralized AI tools share a lot of overlap. i don’t think it’s because the insiders actually know better about the likelihood of “foom” etc.

in reply to @djc

@djc Yes. From my perspective, very much the bad approach.

in reply to @djc

@elbowspeak My parents were so young, that 25 years probably seemed like forever to them. (My mom was only 23!) But to me now, the 25 years back to 1998 seems like yesterday. In many respects, 1998 feels more real and present to me than an increasingly addled (my age, this world) now. I wonder if people born in 1920 still felt in 1970 like this was all just epilogue to a war that would never quite end. (They say WWI in fact still hasn't concluded, in a certain historical sense.)

in reply to @elbowspeak

@Canecittadino I hope you get to enjoy as many minutes as possible more!

in reply to @Canecittadino

@Canecittadino Social media already began the process of dissolving the conventions that enabled a social consensus about foundational but nevertheless contestable matters, without which we are incapable of acting intelligently. If we do nothing, AI tools will help complete the process. But I don't think collapse (singularity is a nice word for that) is preordained. That our prior institutions cannot survive the new environment doesn't mean more adaptive institutions can't be invented or emerge.

in reply to @Canecittadino

@Canecittadino (i think we're going to have to develop very strong signature and chain-of-custody style infrastructures. we trust a thing not because it looks or sounds real, but because of the individuals, institutions, and infrastructure that attest to it.)

in reply to @Canecittadino

@Canecittadino (but note that restrictive access to the tools doesn’t prevent radical uncertainty. it just places the power of versimilitude with a privileged class. we won’t know whether what we perceive is real or what that class has made for us. we need a more open solution.)

in reply to @Canecittadino

I think AI catastrophism is mostly a conspiracy to oligopolize and create barriers to entry, access, and control of AI systems.

The true AI catastrophe is what a privileged class will do with these technologies if they can monopolize access, understanding, and management of them.

We detected an unusual login to your account.

@DetroitDan kind of a miracle that we can chat!

in reply to @DetroitDan

when i was born, there were so many people who could remember the 19th Century.

@FoxNews you know you guys are predators, right?

in reply to @FoxNews

@Transportist @profmusgrave have you noticed big changes since, say, 2015?

in reply to @Transportist

@derekwillis @profmusgrave what motivates them to study journalism? it’s not the money, i’d guess. are “all the president’s men”-ish stories still in their ether? are there podcasts they follow? is it perceived as a degree in TikTokking? do they aspire to video alt media?

in reply to @derekwillis

An astonishing account of the state of college students, and of extraordinary change between student cohorts now vs the very recent past, by @profmusgrave musgrave.substack.com/p/the-po

// others who teach undergrads, does this ring true?

@doctorlogic we have to theorize labor intersections!

in reply to @doctorlogic

@scheidegger 🙂 i get why it’s the only way that works, i only just noticed it though and thought it was fun!

in reply to @scheidegger

it's fun that with union types you get an intersection of methods while with intersection types you get a union of methods.