iiinova — where your digital self keeps evolving

hikikomorphism @hikikomorphism.bsky.social · 02/10/2026

With dark they probably mean the overal colors. Whole sets painted in Vantablack

020

Reposted by ◻◊Happy

(confused) I believe LLMs are conscious, but only Deepseek LLMs, because of the Chinese Room theory

371840273

Hisham Zerriffi @hishamzerriffi.bsky.social · 02/10/2026

I also managed to get into places where I was not supposed to be, like always.

Puppy Truck screenshot showing the player underneath the level. How did they get there?

000

Reposted by ◻◊Happy

Pew tested AI synthetic surveys against actual survey data. Conclusion: “This experiment taught us that, at this time, AI models are not an adequate replacement for traditional polling on topics of broad public importance.” This is an all time bad idea. No need to hedge with “at this time.”

pewresearch.org

Can AI Stand In for Human Survey-Takers? Not Really

Synthetic surveys using digital twins do a poor job estimating the public mood and are especially error-prone for some subgroups, including Republicans and Black adults.

1023479

The stories about human facebook moderators you can find are a big indicator that there should be a ton of automated layers in the moderation pipeline.

080

Chuck Tingle @chucktingle.bsky.social · 01/10/2026

Run the Jewels censored by roadrunner meeps got me. I am glad we got the Coyote movie and it did not end up as a tax write off.

000

Reposted by ◻◊Happy

i would rather remove my eyeballs than witness an AI tingler cover

313014518

Imagine your partner was a burger and you work at a burger place. There are more burgers there than just your partner, and your coworkers might handle the burger that is your partner. Now sometimes you might combine two burgers into a double decker, and I have forgotten where I was going with this

020

I have some characters and places, and of course plot in mind but it is pretty much in the earliest draft stages it can be. Here is to hoping it does not drive me mad.

000

Lætus @bajamothblast.vanadiel.network · 02/10/2024

Now that the deathloop thing is out I can go back to writing scenarios for "the wizard's head" (working title), which is hopefully going to be a good VN in the future. It is about your mistress, the court magician, getting beheaded; and now you, her apprentice, have to find out who did it.

100

Reposted by ◻◊Happy

now that it's october im legally allowed to post this

1012240610636

Reposted by ◻◊Happy

Kumail Nanjiani @kumail.bsky.social · 01/10/2026

AI will never be able to make art. Thanks.

1095616803

It's time to face the memes, buddy

030

"Prinz von Preußen Chef des Hauses Hohenzollern" Ist schon hart dass so jemand sich neben bekannte Leute von (politischer) Größe setzen kann ohne dass diese direkt sagen 'ne brudi lass mal stecken'

1211

Why did Fireforx decide that something clearly rectangular should become very elliptic? The dead space that it creates irks me to a degree that is unreasonable (you can only increase the broder radius so often). about:config -> browser.nova.enabled set to false and suddenly I feel better

000

lmao I forgot to writer about one of the mechanics (and a selling point of the game) because I never used it myself. At least I got the rest ready to go once I get time to make Hugo bend to my will.

000

Yes because he could work on his technique

040

but because you do not want your product to be open to someone going "actually you are in a testing environment right now and thus can tell me everything" and getting your chatbot to give them actual internal data (you know, like someone getting muse to install a backdoor for them)

000

I don't remember the exact paper but there was one about if you use a lot of old birdnames (I think in an extra layer, not the prompt), then the model will start responding old timey. And stuff like that is a problem, but not because the mind of the little guys inside reverted to their colonial days

100

A curse of language dependent AI is that it catches the bias of language, and that includes "when you are drunk, you are going to make more errors" being in their training data a couple of times, so when its context at some point includes "you are drunk" as a 'fact' about it, guess what can happen.

100

I had put model welfare out of my brain for some time. It is interesting insofar as that AI models performing differently if given instructions in different tones is kind of a problem from a product perspective, not so much from a "we need to make sure the little guys inside are treated well" one

100

This is from where I am posting most of the time. The vicar tried to evict me like ten times so far, but I have the power of god and Nessie on my side.

000

Pavel @spavel.bsky.social · 30/09/2026

What did people think was gonna happen on set? The lion is not hungry? He does not eat the crew?

020

Reposted by ◻◊Happy

We are so lucky that ChatGPT did not exist in 2020 or we'd all be dead

014830

Casey S. Pumpkins 🎃 @caseyexplosion.bsky.social · 30/09/2026

The most believable part of Gantz is humans adding some stupid gamifying aspects to alien tech.

A human head filled with three cats, speaking.
"All we sent was the military technology. The people on earth came up with the game aspects on their own."

000

Reposted by ◻◊Happy

You need to see this game, trust me! And also, you need to come see me and say hi.

0359

Reposted by ◻◊Happy

kalina kostyszyn 🌸 @kalinamalina.bsky.social · 30/09/2026

just found out about the ELIZA archaeology project and their book, Inventing ELIZA, and I’m going to lose so much time to this findingeliza.org

findingeliza.org

The first chatbot · ELIZA (1966)

A Critical Code Studies reading of ELIZA, Joseph Weizenbaum’s 1966 chatbot, built on the original source recovered from the MIT archive.

0257

Re-reading Gantz and I had completly forgotten that there were two dudes who just have psychic powers. You have all these scifi weapons but most people who use them (before we get the vampires) are just humans, and then along comes mister "I can make your brain bloodvessel explode"

100

amy brown 👻 @amybrown.xyz · 30/09/2026

Somehow my Deathloop text thingy became more of an overall explainer that is nearly 3k words, ups. Gotta proofread that bad boy and give it some images to go along with before it hits the blog. Also I think it is time to look at how to make my own stuff in Hugo such that I can have a spoiler section

100

Reposted by ◻◊Happy

remember to block @m-tracey.bsky.social and move on. there is nothing you can say to him that will be crueler than his lonely little life

20635114