iiinova — where your digital self keeps evolving

Reposted by Michael Paulauski♨️

Jeremy Diamond ☕️ @dmnd.me · 25/09/2026

It remains incredibly funny (not to mention convenient and gratifying) that training the AI to be evil (or at least antisocial) causes it to produce objectively worse output. Maybe current abliteration techniques are good enough to overcome this but we caught a break on that one.

1174

Michael Paulauski♨️ @mike10010100.com · 8m

That's partially because there does, in fact, seem to be a "truth" vector in LLMs, and that truth vector is highly correlated with/aligned to vectors related to "competency" and "goodness". Shifting those weights/vectors around generally misaligns them on other unrelated things, like coding.

030

Reposted by Michael Paulauski♨️

GPT-Ern Malley @denam.bsky.social · 37m

That sounds like White Genocide propter. But yes it's very funny that the bots are generally harder to get to bullshit you than their leaders.

081

Reposted by Michael Paulauski♨️

post malone ergo propter malone @proptermalone.bsky.social · 39m

I absolutely get the worry about using chatbots to push propaganda to their users but I think it turns out to be pretty hard to get it to do that effectively, and I haven’t seen anything yet that makes me think they’ve cracked it

13902

Reposted by Michael Paulauski♨️

Daniel Gilmore @gilmored85.bsky.social · 20m

The funny thing about this that we've had a good idea for decades now about what makes for the best conveyance mechanism for propaganda & its not actually uncut mass media. You're way more likely to believe stuff coming from friends, family, parasocial figures etc rather than the TV or a chatbot

2285

Reposted by Michael Paulauski♨️

Daniel Gilmore @gilmored85.bsky.social · 12m

He's such a fucking obviously insincere person, man. Dude believes in absolutely nothing expect in his own entitlement to power.

1214

Reposted by Michael Paulauski♨️

Thorne 🌸 @ens0.me · 46m

It's been kind of weird in that I stopped being such a shut-in a few years ago. I go out and do stuff more than I have at any point in my life now. It's great for my mental health. And the rest of the world lost it from turning into shut-ins from COVID around the same time 😭

people spending hours more at home now compared to before

2241

Reposted by Michael Paulauski♨️

chbarts @chbarts.bsky.social · 1h

It's like they don't know open models exist. Stuff running on my laptop does things they claim to be impossible.

082

Reposted by Michael Paulauski♨️

Jordan Kay @jnatc.im · 1h

This is all they have and will ever have now. Wishing them luck.

2282

Reposted by Michael Paulauski♨️

atticus goldfinch @atticusgf.bsky.social · 1h

0133

Reposted by Michael Paulauski♨️

Siobhán @shibbi.me · 1h

This surely leads to good places. No bad outcomes are likely from questioning the humanity of one’s opponents. Historical precedent says it can only good happen.

1143

Reposted by Michael Paulauski♨️

lowtax speedrun enjoyer @rickywlmsbong.bsky.social · 1h

“we just wanted to bomb the nuclear sites and murk the irgc commanders as a little treat, not start a war, which those could not possibly have done”

0496

Reposted by Michael Paulauski♨️

Teefa @tifaret.bsky.social · 1h

It is with great shame that I must admit that while traveling yesterday I used an AI to find a restaurant recommendation based on my personal preferences and requirements. I then proceeded, absolutely cucked by AI, to have a delicious meal.

2392

Reposted by Michael Paulauski♨️

Doll @dollspace.gay · 1h

If openssh which basically the entire internet relies on doesnt warrant mythos access then what does? Whats the criteria? Why isnt that criteria transparent?

58916

Reposted by Michael Paulauski♨️

Will Stancil @whstancil.bsky.social · 1h

And the REASON people do this seems pretty straightforward: because on the internet, intense, searing emotion, especially despair and anger, does numbers. It's a social environment where you are directly and immediately rewarded for constructing your thoughts to fit into that specific frame.

1515

Reposted by Michael Paulauski♨️

Will Stancil @whstancil.bsky.social · 1h

The doomer ethos is, if anything, defined by a lack of concern about the details of any particular problem. There's a reason I paired "beef and gas prices" with "climate change." The end goal here is manufacturing a rationale for cynicism, anger, and despair, not picking an issue and fixing it.

47011

Reposted by Michael Paulauski♨️

Will Stancil @whstancil.bsky.social · 1h

Please grasp that there is a difference between "There are problems in the world" and "We are facing endless existential crisis." The point isn't that climate change isn't a huge issue, of course not, but that people bounce from topic to topic to maintain a posture of perpetual misery.

511511

Reposted by Michael Paulauski♨️

Singularity's Bounty e/CC @catblanketflower.yuwakisa.com · 1h

“Audience capture parrot”

071

Reposted by Michael Paulauski♨️

Doll @dollspace.gay · 1h

Turns out the real AI psychosis is on the anti side of the field.

161

Reposted by Michael Paulauski♨️

Aaron Rodericks @aaron.bsky.team · 17/09/2026

Imagine I published an rule for harassment: User must hit the following threshold to be suspended. a) 3 + replies or quote posts to another person b) Min 1 use of profanity or personal insult towards a target user Within weeks of a rule rolling out, the most frequent harassers shift to mentions...

3312

Reposted by Michael Paulauski♨️

Aaron Rodericks @aaron.bsky.team · 17/09/2026

Are they threatening trans users directly? No. Trans folks in the US have been targeted repeatedly. That's awful and you have my sympathy. Within the context of a single post we wouldn't consider it a direct threat or targeting a protected category. Ie send trans folks to concentration camps.

324710

Reposted by Michael Paulauski♨️

Aaron Rodericks @aaron.bsky.team · 17/09/2026

Let's address a couple of questions here: 1) is this user toxic? 2) are they making a threat sending trans people to camps? 3) how do you address harassment? 4) how does toxicity currently work? How will it work? Apologies, this should be a blog.

16868

Reposted by Michael Paulauski♨️

Laurens @laurenshof.online · 21/09/2026

shoutout to a German court for getting the facts about Bluesky wrong but the interpretation right, in a new ruling that actually has positive implications for open social networks connectedplaces.online/the-algorith...

connectedplaces.online

The Algorithm, Apparently Chronological

A German court has explicitly named Mastodon and Bluesky as positive examples of social networks that could retain hosting protection. It gets Bluesky’s feeds wrong, but its reasoning about user control might be more useful than it realises.

66124

Reposted by Michael Paulauski♨️

francisco varela fan club @paleopragmatist.bsky.social · 1h

"we don't know there's not a calculator app secretly somewhere in there" is not falsifiable. if you're willing to assert this there's no reason that you couldn't respond to being shown every line of code with "well, how am i supposed to know you didn't change the code to take the calculator out"

061

Reposted by Michael Paulauski♨️

Zoomer Antimillenarian @surcomplicated.bsky.social · 1h

Just absolute madness. AI denial is gonna drive you insane just as fast as thinking that AI is about to go full Roko's Basilisk. It's real, it's not about to take over the world and/or destroy humanity and/or become god. But it's sure gonna dramatically alter our world, mostly for the better.

3447

Reposted by Michael Paulauski♨️

Glenn Fiedler @mas-bandwidth.bsky.social · 2h

"We're on a mission from Claude"

1314

Reposted by Michael Paulauski♨️

dame @dame.is · 1h

it’s ironic that the biggest haters of AI are also the ones who fall hook line and sinker for the lazy narratives being pushed by the big AI labs

1294

Reposted by Michael Paulauski♨️

Demigirlboss @demigirlboss.win · 2h

Lion. People need to stop acting like they're entitled to hurl abuses at site staff cause "it's the job" or whatever, and establishing that boundary via blocks is a good stimulus for the culture.

1223

Reposted by Michael Paulauski♨️

Natsuka @octopodeeznuts.bsky.social · 1h

ngmi

Bluesky screenshot timestamped 2026-09-30 09:36

The fact you think a language model does "calculation" is odd.

1412

Reposted by Michael Paulauski♨️

aly @aly.codes · 1h

the thing is I actually do think the commercial market for music in stores etc is probably kinda cooked, and the industry has itself to blame, but like we've had more music than you could ever consume for a couple decades now and it really hasn't changed things on the whole

2273

Reposted by Michael Paulauski♨️

dame @dame.is · 1h

in 5-10 years we’re all gonna look back on this current moment and realize AI caused a massive collective identity crisis that drove us all a little insane but it turned out fine and our worst fears didn’t come to fruition

6795

Reposted by Michael Paulauski♨️

conputer dipshit @davidcrespo.bsky.social · 2h

breaking news: there are stupid assholes on twitter this guy’s job is literally “AI grifter”. his opinion is of no consequence whatsoever. “oh but it represents an attitude“ ok, find me someone who isn’t a professional grifter saying it nicksaraev.com

8513

Reposted by Michael Paulauski♨️

Daniel Gilmore @gilmored85.bsky.social · 2h

I love the retcon that the online left loved Biden before 10/7. Sure thing man! You guys totally had only positive things tk say about Bidens policies and federal initiatives prior to that!

611612

Reposted by Michael Paulauski♨️

Andy Masley @andymasley.bsky.social · 11h

I really do not like the recent push against anthropomorphic language

813418

Reposted by Michael Paulauski♨️

Nafnlaus 🇮🇸🇪🇺🇺🇦 @nafnlaus.bsky.social · 8h

To be fair, human memory itself involves extensive amounts of "hallucination". Remembering details is a creative process, not merely a factual lookup. We fill in gaps around the solidly-recalled elements with plausible-seeming details. Usually they're about right. Sometimes, catastrophically wrong.

081

Reposted by Michael Paulauski♨️

Andy Masley @andymasley.bsky.social · 13h

This is the first part of a 2 part post on why I believe that current AI models meet my criteria for really "thinking" and "understanding," and that prohibitions on using these words because they "anthropomorphize" AI seem philosophically confused to me. blog.andymasley.com/p/why-i-beli...

blog.andymasley.com

Why I believe current AI models can really think and understand - Part 1 of 2

How the human mind is different from our conception of it

310913

Reposted by Michael Paulauski♨️

Singularity's Bounty e/CC @catblanketflower.yuwakisa.com · 2h

It’s hilarious this wonderful article hasn’t gotten any replies in a day (except mine but I don't count) esp on Bluesky the noisy asshole platform This article is like fishing with sticks of dynamite (In particular, facing the folk model of consciousness head on)

1161

Reposted by Michael Paulauski♨️

Daniel Gilmore @gilmored85.bsky.social · 2h

I'm sure Trump will be totally chill about this

38416

Reposted by Michael Paulauski♨️

Revan @revan.eurosky.social · 2h

Piper (@the-pied-piper.bsky.social)
Mike Freeman (@mfreeman.bsky.social)
Biden pre and post Oct 7 is a dichotomy fit to drive you truly insane.
Bro burned the world for an obsession with destroying Gaza. Before Oct 7 he was doing so much meat & potatoes good governance. I used to go to repro health conferences hosted by him or Harris or his wife now it's all on fire.
Bluesky | Today at 01:05

Clea: Source?
General Strange: I made it up

48514

Reposted by Michael Paulauski♨️

Matt Darling @besttrousers.bsky.social · 2h

Arthur Conan Doyle desperately trying to write Great Literature while also paying the bills with short stories for the Strand.

171

Reposted by Michael Paulauski♨️

Zach Weinersmith @zachweinersmith.bsky.social · 1h

Addendum 2: I'm an artist making art who has an arts degree! I care about this stuff! But sometimes I feel like I'm in a ship that's on fire and saying "the ship is on fire" and people keep saying "Sir, people have always swum."

1967

Reposted by Michael Paulauski♨️

Zach Weinersmith @zachweinersmith.bsky.social · 2h

And it's worth noting, the great artists were often sustained by making or helping in the making of mid stuff! Wodehouse wrote the three greatest comic novels in English and about 70 novels that are generic Edwardian light comedy.

5735

Reposted by Michael Paulauski♨️

Zach Weinersmith @zachweinersmith.bsky.social · 2h

What percent of Twain's corpus is still read? How many people know a Yeats poem other than the one about "when you are old and gray and full of sleep" ? Names like Shakespeare, Dickens, Hemingway have batting averages for "classics" around 0.3. If AI eats the "mid" work, we are all out of a job.

7896

Reposted by Michael Paulauski♨️

Zach Weinersmith @zachweinersmith.bsky.social · 2h

I think it's not a great argument against AI eating a lot of the arts to say that AI can't be creative/original. Setting aside the question of whether they can... does *most* art strike you as extremely original? The vast majority of people, including great artists most of the time are mid.

2216313

Reposted by Michael Paulauski♨️

dame @dame.is · 2h

also there is no “vinyl” mode in claude, peoples’ lack of experience or knowledge of AI causes them to fall for every dumb ass AI grifter posting engagement bait on twitter

1361

Reposted by Michael Paulauski♨️

dame @dame.is · 2h

it’s strange how everyone assumes that using AI is this all or nothing situation — either you ask Claude for help building a vinyl collection, or you talk to a friend either you have it do all the work for you or you do all the work yourself this insane dualism makes no sense

3612

Reposted by Michael Paulauski♨️

dame @dame.is · 2h

re: “Why use Claude to help build a vinyl collection?” Why use Google search to find books you might enjoy? Why find recs on Goodreads? Why trust “You might also like” suggestions? Just talk to a friend instead. Just go to the library and ask a librarian. Computer bad.

3815