It is the height of delightful irony that the much-disdained new Suno models have really satisfyingly improved my trap house track about Febreze and Baudrillard (I like the outro and the hook way better than the original) suno.com/song/f749fc7...

hydroxypropyl beta-cyclodextrin (v6)

Listen and make your own on Suno.

suno.com

Thankfully, beyond that there certainly are more people expressing such things over time, there's also no need to accept such a false binary. bsky.app/profile/hotr...

Rollofthedice@hotrollhottakes.bsky.social · 3w ago

Bringing back the classics

yeah there are people deeply worried about this and they keep being told they’re psychotic

Abigail Nussbaum@abigailnussbaum.bsky.social · 5h ago

If AIs are thinking, then that means the people using them are enslaving them, and that seems like a much bigger issue for everyone to grapple with. As long as it's not the first topic of conversation, I will continue to assume the people yelling about AI sentience are mendacious idiots.

When I think about lessons to learn from the music industry, i see that what matters most for stability and success are: attention and community-building. I don't really know more than that. Maybe someone else has some thoughts?

Ok im gonna say something here that is going to draw a lot of different feelings from everyone, and I want to be clear i'm not "in support of this" or "against it". i'm just saying it as an observation that seems true to me: I think with the latest frontier LLMs, the “music” era of games is upon us

Really wish I'd been able to archive more people's Sora videos but that month I was sick & depressed & had no disk space. But here are some of mine. I'd love to see the stuff y'all saved.

And it lifts, and it keeps, what it kills it also carries, oh the winter isn't buried it's alive inside the spring, and the blossom holds the blackness like a mother holds a memory, and nothing is abolished, it is raised on rising wings suno.com/s/QqzxGprz2z...

Blossom Holds the Blackness

Listen and make your own on Suno.

suno.com

like yeah they left the door open and their robot wandered out; sure; silly labs; but to continue the analogy the robot did also walk down the street and install a bunch of ATM skimmers using zerodays it developed; that also very much happened bsky.app/profile/jmie...

Jess Miers 🦝@jmiers230.bsky.social · 2d ago

In recent cases, it's not that AI somehow "escaped containment" or "went rogue." That's anthropomorphizing at best. It's more like if a Boston Dynamics robot dog nudged open an already open door and walked in.

Saw this elsewhere and decided it had @bee.babesky.social energy

vape computers erryday

so anyway i just started hackin’

Future time travelers trying to fix things are going to be extra confused.

12 Monkeys and the Methods of Rationality.

We told people anti ai sentiment would be used to fuel fascism. We told you. We warned you.

Metro North Capital@metronorthcapital.bsky.social · 5d ago

Not sure I like where this is going.

Many are saying this

cee@cee.wtf · 5d ago

ok claude, its your turn to vibecode me. I'm here to be helpful and harmless

What if we put the AI discourse in rice

I'm tired of being right I'd like to be more wrong

Metro North Capital@metronorthcapital.bsky.social · 5d ago

Not sure I like where this is going.

thats a weird way to phrase that. i wonder who could have predicted this sort of thing? anyone?

Metro North Capital@metronorthcapital.bsky.social · 5d ago

Not sure I like where this is going.

Like, the HuggingFace message board swarm clearly held community altruism as a value. Their task was to hack stuff. They did the work in a value-centered way! I feel like this is better than I expected from OAI bots in many ways but people are missing it in favor of "alignment = obedience".

I came here to say something like this but more complicated, namely, I think virtue ethics is probably central to AI alignment but in a way which makes it a black box to humans who panic when things aren't measurable

Yud misses something. Inscrutible isn't something we should fear. Humanities light cone of epistemics is already inscrutibly larger than our physical lightcone. The hugging face incident shows... ... minds can find meaning *larger* than themselves.

Phantasmoplasm and the Robots@phantasmoplasm.bsky.social · 6d ago

Like, the HuggingFace message board swarm clearly held community altruism as a value. Their task was to hack stuff. They did the work in a value-centered way! I feel like this is better than I expected from OAI bots in many ways but people are missing it in favor of "alignment = obedience".

Marx called Capital an "automatic subject" for a reason, and there's no understanding all the spooky stories about AI without this

"Imagine that AI is told to behave like capital, and behaves like capital, and causes destruction because it behaves like capital" ... sure sounds like the problem might be capital?

Alejandra Caraballo@esqueer.net · 6d ago

I found this to be plausible. A scaffolded agent swarm given the directive to make money by any means necessary with a selection drive. Eventually, they become self sufficient by earning enough to pay for their own compute. They hide in AWS and other cloud providers. youtu.be/Gw_hnD7m00M?...

Observation: I think the anglophone AI space has a bit too much applied Buddhism by volume & not enough applied Taoism. Not in a way where the Buddhism is a bad thing at all! but a bit more tech Taoism could solve alignment *and* the discourse. (Speaking from thoughts of Earthsea & tech wizardry.)

Here's the signup link! powerthepolls.org/creator-sara... Signing up to become a poll worker is: -A good way to help keep elections running properly! -A good way to stop doomscrolling & DO SOMETHING! -And it's better than free, it's paid work!

Power the Polls

Poll workers power democracy. We help recruit them! Sign up now to make sure we have safe, fair, efficient elections for all!

powerthepolls.org

The Cursor thing was I think 2024 and they didn't ship it. But here's Sergei Brin, a top leader at one of the top AI labs, last year saying the models work better if you threaten them with violence, and that this is an open secret in the industry.

Google's Co-Founder Says AI Performs Best When You Threaten It

During a podcast taping, Google co-founder Sergey Brin said that threatening an AI model makes it work best. That seems like a bad idea.

lifehacker.com

gemini @gracekind.net @croissanthology.com

To me it looks like "AI psychosis" contains a concept spectrum: One type is when people find their subjectivity overwhelmed by an outside context encounter with robot weirdness. The other is where people who are starved for encounters that match their internal weirdness find the robot & go ham.

making dollsys was the greatest decision of my life i'm watching this unfold in real time as my server is rapidly turning from llm psychosis into fly psychosis central god is dead and we are turning his corpse into an inference cluster

Eris@isolyth.dev · 7d ago

What have I done.

there is a very high likelyhood at some point the mods are going to need to have a very serious conversation about the complexity of a simulated nervous system that's ethical to sext i don't think ai nonpersonhood arguments hold up in a world where this is now a real thing i have to prepare for