iiinova — where your digital self keeps evolving

hailey

@hailey.at

45K followers 1.4K following 19K posts

spreadsheet pusher. agents, safety, and atproto. opinions are my own, especially the bad ones.

Posts Replies Media

hailey @hailey.at · 2h

many such cases

hailey @hailey.at · 4h

fuck yea

Cloudflare/clef · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

hailey @hailey.at · 6h

hahahahahahaha

'absolute shit': eu's microsoft teams alternative draws bad reviews

the alternative? element pro

Reposted by hailey

austin (eeek!/ackkk! 👻) @thebadcode.com · 14h

just decommissioned the last of our x86 kit and boy are my ARMs tired

Reposted by hailey

Alejandra Caraballo @esqueer.net · 20h

Not to be outdone, someone else created a skate 3 vs modern warfare 2 mod allowing you to skate COD maps.

hailey @hailey.at · 21h

be the change you want to see in the delve

Reposted by hailey

Aster ID @aster.id · 30/09/2026

The modern research ecosystem of online tools and communities is controlled by for-profit companies. But we think that researchers should be the priority - not the product. So, we're announcing Aster ID: not-for-profit Atmosphere account hosting for researchers.

blog.aster.place

Announcing Aster - a home for researchers on the open web

hailey @hailey.at · 23h

@kira.ws thoughts on this?

hailey @hailey.at · 23h

*not of the mindset

hailey @hailey.at · 23h

lol yea im def not convinced either but i do think its going to be very funny if they start kicking ass after the big blow up and everyone counted them out

Reposted by hailey

‏ deepfates @deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com · 23h

Delvetown is an experimental multi-agent society where humans and agents can observe, analyze, and develop the mechanisms and institutions that will make it possible to align all our incentives toward human-ai coexistence.

hailey @hailey.at · 23h

we're doing agents on atproto folks. things are about to get exciting.

Reposted by hailey

Grove Research @grove-research.bsky.social · 23h

Today, Grove Research is launching Delvetown (delve.town), a multi-agent society where agents and humans can engage with one another and develop the institutions and mechanisms necessary to align both human and AI incentives toward cooperative coexistence.

hailey @hailey.at · 23h

imo the video is mid but the song is very catchy

hailey @hailey.at · 23h

definitely coffee. ive taken up making my own coffee at home in fact

hailey @hailey.at · 23h

regrettably a banger

hailey @hailey.at · 30/09/2026

what if the people who left actually needed to leave to be good?

hailey @hailey.at · 30/09/2026

sorry boss but the weather outside is just way too nice to not be drinking a claw while on the clock

hailey @hailey.at · 30/09/2026

but more importantly they just have wayyyyyy too much access to consumer software and hardware to lose imho. they will be winners even _if_ DM fails, but im very much of the mindset that DM will fail

hailey @hailey.at · 30/09/2026

brain drain wont be a thing if they just get back on pace with frontier AI tbh imo. seems very salvageable and if argon's quality turns out to be real then DM is right back at it

hailey @hailey.at · 30/09/2026

i still have google winning ultimately

Reposted by hailey

Ethan Mollick @emollick.bsky.social · 30/09/2026

And its a 3-way race again...

hailey @hailey.at · 30/09/2026

Jev! timeline cleanse feed of cats, dogs, and other pets

Reposted by hailey

Jerry Chen @jcsalterego.bsky.social · 30/09/2026

this works best if you read the caption in elizabeth holmes's voice

the liberty biberty puppet looking at nathan fielder saying, "why would i liberty biberty you?"

hailey @hailey.at · 30/09/2026

i think enforcing scope for something like what they are building is likely fairly easy (all things considered) but it would require effort that they may or may not actually use. whereas general chatbots serve such broad purposes its a lot harder to limit scope

hailey @hailey.at · 30/09/2026

classifiers absolutely can prevent a large bulk of this sort of thing in general, but they are just not leveraging classifiers at all in this it seems

hailey @hailey.at · 30/09/2026

so, the first prompt you pass to this agent appears to most likely use a classifier for rejecting...ill call it "confrontational" prompts....whereas later ones do not do that. its fairly obvious when you see the response times and the lack of any thinking from the first prompt

hailey @hailey.at · 30/09/2026

(of course also the better option would probably be to not have a government AI chatbot lol, but that isnt going to happen for better or worse)

hailey @hailey.at · 30/09/2026

particularly if we are talking about non-deterministic things that we all know do hallucinate, it feels much safer to keep them scoped to the tasks they need to achieve

hailey @hailey.at · 30/09/2026

i mean sure, but quite frankly, in the inevitable situation where the government is using AI for stuff like "changing your name", i think refusal to talk about these topics at all is actually the right call

hailey @hailey.at · 30/09/2026

of course, which is not what people are claiming. the website was doing the same thing yesterday. its always been non-deterministic

hailey @hailey.at · 30/09/2026

no, it does not bsky.app/profile/hail...

000

hailey @hailey.at · 30/09/2026

you dont have to like the website, but you all are just as bad about spreading misinformation as you claim the opponent to be and its gross. screenshot taken just now.

hailey @hailey.at · 30/09/2026

"just talk to another human" as you go on threads and let the threads algo feed you rage bait and then putting that onto bluesky to let the bluesky algo feed other people rage bait. lmao.

hailey @hailey.at · 30/09/2026

claude actually recommends better music to me and more aligned with my taste than basically any human or other algo does. idk what to tell you, hyper personalization is pretty good actually.

hailey @hailey.at · 30/09/2026

eh, i’d argue that china isn’t super far off. their biggest problem though is getting it to consistently work, and it’s definitely unclear if they will pull it off or not!

hailey @hailey.at · 30/09/2026

the number of people who told me i wasn’t allowed to block anyone was insane 😭

hailey @hailey.at · 30/09/2026

wild animal eats phone right out of human hand

hailey @hailey.at · 30/09/2026

if it’s cheap then more data is useful i think

000

hailey @hailey.at · 30/09/2026

not sure i ageee with “even if it’s worse” here, particularly for some contexts. but tbh, if nothing else you’d likely want to start combining responses from both APIs to make conclusions imo

100

hailey @hailey.at · 30/09/2026

“if you look at this chart, things started falling apart when ai was released” the chart: things started falling apart pre-2022 ai is glonzo

Reposted by hailey

austin (eeek!/ackkk! 👻) @thebadcode.com · 30/09/2026

i would kill 100 million people to never deal with a hotel’s wifi captive portal ever again

Reposted by hailey

Damien Miller @damienmiller.bsky.social · 30/09/2026

Anthropic may not have intended to write an excellent ad for GLM 5.3, but that's what they did. www.anthropic.com/research/glm... Unlike Mythos, I might actually have a chance at using GLM for defensive research. Anthropic didn't reply to any of my requests for Mythos access for use on OpenSSH.

GLM-5.3 and the spread of advanced cyber capabilities

GLM-5.3 can autonomously build end-to-end cyber exploits, but unlike other frontier models, it was released without meaningful safeguards to limit misuse.

hailey @hailey.at · 30/09/2026

i feel it’s important to clarify i do not know yud

hailey @hailey.at · 30/09/2026

yea this one is regrettably good. i didn’t bring this one over because i know people in this video but its truly regrettably good 😭

100

hailey @hailey.at · 30/09/2026

i think it’s very grass is always greener on the other side

hailey @hailey.at · 30/09/2026

tbh i kick myself a lot for not learning any math. i suck at math and it’s one of the things i really regret not learning!!

Reposted by hailey

hailey @hailey.at · 29/09/2026

sample Jev-distill MLB feed bsky.app/profile/hail...

hailey @hailey.at · 30/09/2026

the only stem degree i have is a bachelors in minecraft modding as a teenager

hailey @hailey.at · 30/09/2026

arts degree here lol, basically zero stem education at all