iiinova — where your digital self keeps evolving

kubie (soweli Kupi) @fox.northsky.social · 6h

guy who’s only seen the RAADS-R test and thinks autistic people must all watch a lot of television

000 0

kubie (soweli Kupi) @fox.northsky.social · 6h

finally. something to bring the country together

020 0

kubie (soweli Kupi) @fox.northsky.social · 16h

the first indictment happened while I was an undergrad at Stanford. US v. Holmes happened right when I was raising my first venture round. i genuinely hate everything about VC capitalism now but this film is my superbowl

010 0

kubie (soweli Kupi) @fox.northsky.social · 16h

it genuinely falls like right on the line of ‘actual reply that could happen’ / ‘incredibly dry satire’ and I don’t know how to tell

020 0

Reposted by kubie (soweli Kupi)

kat claws @kat.meangirls.online · 08/09/2026

showing up to the theater with my girlfriends and we're all wearing a black turtleneck

1233 0

kubie (soweli Kupi) @fox.northsky.social · 16h

the vibe I get from this answer is definitely “this person is so free and normal”

110 0

kubie (soweli Kupi) @fox.northsky.social · 20h

if my own experience with Scouting holds, they will all get D cups and become U-Haul lesbians by 2035, which could be fantastic for unit cohesion

050 0

kubie (soweli Kupi) @fox.northsky.social · 21h

glad to see the UK is finally taking my idea seriously: what if we did Boy Scouts, but with 70 year old men, for fascism

130 0

kubie (soweli Kupi) @fox.northsky.social · 16/10/2026

I love self-help articles that just point to a diagnosis. seeing a new set of 3-5 capital letters in a row really fixes things for me in a way actual advice could never.

0100 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

learning to dj solely so I can play 4k carpet so loud I break the speakers in the club. then I retire in obscurity

040 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

wait really? i assumed he had to be tenured from reading nlab

220 0

Reposted by kubie (soweli Kupi)

🇵🇸Tim Henke (tɪm 'ɦɛŋ.kə) @timhenke.bsky.social · 10/09/2026

Prompted by the scandal around Navier-Stokes, Andreas Thom raises the possibility that OpenAI may also have stolen the work on non-sofic groups:

1/3 I want to record a number of points that I realized after reading up on the controversy. In fact it recalls an exchange I had with OpenAI after its non-sofic-group announcement. It exposes a serious, unanswered question about whether researchers can trust OpenAI with unpublished mathematics.

I wrote in an email to Mark Sellke and Sebastien Bubeck shortly after the surprising finding of a non-sofic group using the methods of Kun and myself:
“Another point is that I and a colleague in Dresden were discussing the expander matching problem and various extensions of the work with Gabor Kun actively over the last months with ChatGPT, so that we are of course curious if that was part of the training data or accessible to the reasoning process. There is a certain (frankly unacceptable) lack of transparency here; and I fear it will damage the communal process of math more than the new AI-generated results will benefit the subject.”

Mark Sellke’s complete answer was: “Regarding your conversations with ChatGPT: that did not happen.”

I had explicitly asked about two different things: (1) whether our conversations entered training data, and (2) whether they were accessible to the solving process. The categorical answer now looks as though it addressed only direct access under (2). No such qualification, explanation, or evidence was given. I take this as dishonesty to say the least.
OpenAI now says in the Buckmaster-Alpöge case that no specific user data was accessed, but adds that it “cannot rule out that de-identified data derived from their usage of our products helped improve our models.”

2/3 OpenAI is drawing a distinction that its answer to me erased, despite the fact that my question explicitly made that distinction.
We are not required to reverse-engineer OpenAI’s internal training pipeline to establish what happened. Only OpenAI has the relevant data for that. If Sellke and Bubeck give a categorical denial, it must disclose the basis: product and privacy settings, relevant datasets and checkpoints, and what “de-identified data derived from usage” means.
I disabled model training on 29 June. That control is still only a promise whose implementation users cannot audit, and it is prospective: it does not answer what happened to earlier conversations or to derivatives already selected. OpenAI’s answer did not mention the setting or any account-specific check.
There is also a mathematical parallel, though a less explicit one. The approach of Kun and myself was not the main line of attack on non-soficity, in fact there were other more promising approaches along the line of quantum games etc. OpenAI’s detailed command of our techniques therefore made me wonder how the model found this route. If nonpublic research supplied by users improved a model and the provider then used that model to race those users to publication - without informed consent, disclosure, or credit - that would be ethically indefensible. De-identification may remove a name; it does not remove the intellectual content of a mathematical idea. Sellke and Bubeck seem to be blind to this simple moral aspect.

3/3 Sellke’s categorical answer was, at minimum, unjustifiably broad and materially misleading; looking back it was plainly dishonest. Bubeck’s acknowledged attacks and his attempt to exclude Alpöge from a proposed OpenAI-proof paper deepen the concern that there is a loss of moral compass. Taken together, this raises serious questions about their judgment and personal integrity.

113750 0

Reposted by kubie (soweli Kupi)

👁️🍄🪸 @bloomfilters.plus · 11/09/2026

the mathematics award winning fields medalist and known himedanshi yu deng just said that if ai puts all mathematicians out of a job he's gonna go home and write yuri novels. girl please do it anyway i need to see what yuri a fields medalist mathematician can do omfg

212475626 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

thank you so much ^^ that genuinely means a lot, especially coming from you and the remarkable exposition I've gotten to see in your work :)

010 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

when the field of AI has achieved its new goals, we will see amazing feats, entirely centralized in the hands of billion-dollar companies, to leverage as a cudgel and a permission structure to extract even more from us. this isn't the future I want, and so--to me--the field is already dead.

040 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

I guess what I'm trying to say is, capitalism has _always_ made academic fields brittle and money-seeking, and the already questionable undercurrents of attitudes in AI research a decade ago made it even more susceptible to cultural takeover in a way that was symbiotic with capitalistic co-optation

141 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

and all of this has only centralized AI research, incentivized work on a single paradigm, and made the field foundationally dependent on literally-genocidal hyperscalers rather than anything that intentionally builds toward a future where we've moved past them.

140 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

...and it doesn't take much review of what NeurIPS papers get published over the last 3 years to understand there's a trend in what gets funded and what doesn't. we are seeing a generation of researchers pour efforts into LLM prompting, LLM fine-tuning, LLM applications, LLM everything.

130 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

this isn't to say the whole field is currently nonsense, or that there's nothing of value. I've seem a lot of beautiful work still come out. but it's often a lot cheaper and faster to make a paper about improved prompting of LLM agents, or fine-tuning LLM models...

120 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

we are seeing what started as an academically-focused field gradually erode into a theocratic, capital-driven wasteland. and I'll admit, it still makes me sad. I used to keep hope there could be something beautiful there.

151 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

the publish or perish death spiral has gotten worse. focus has shifted more from the computationally interesting to the commercially viable. TESCREAL nonsense mixes in with actual research. a millenialist vision of the future and socially-driven "P(doom)" beliefs have muddied the water of reality

121 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

I bailed to get a math degree, and since I've seen the boat just continue to sink in VC money. a field which could have brought us something beautiful and interesting choked by an artifical abundance of computing power and streamlined abuse of human rights to bootstrap datasets and models.

130 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

the boat was sinking then. I spoke with Andrej Karpathy after one lecture about how T*sla handled collected data, and it just... it made me feel bored and like the field was going nowhere. and I don't think I was wrong.

130 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

but I remember studying these things at stanford in 2017-2019 and watching the field fall apart I remember seeing an increasing flow of garbage papers come out of Stanford Vision Lab with quality of datasets sinking. one summer I published helping show one was worse than a naive averaging strategy.

150 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

there is a feasible world where we built something more beautiful, something actually efficient, something with decent inductive biases. computational chemistry was headed that way with interesting GNN architectures ca ~2015 and there's still echoes of a lot of that work.

162 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

there is nothing feedback loop-y or virtuous about pulling words out of your sleeve like a clown with a line of stochastic handkerchiefs nature played a cruel trick in making the most "good-enough" scalable system the one which is most computationally wasteful and structurally uninteresting

262 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

cybernetics is what hofstatder did, not hinton, and i will die on this hill

160 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

will say that the PM2.5 here is comparable to where I was in the US though, and it seems like that’s what’s failing WHO targets.

020 0

kubie (soweli Kupi) @fox.northsky.social · 14/10/2026

saw a local representative advocate for curtailing “dangerously high air pollution” here. the AQI: 2. not 20, not 200, just… 2. there are little bits of europe i kinda love ngl

130 0

kubie (soweli Kupi) @fox.northsky.social · 13/10/2026

on the plus side, i made it. yaaaaay

040 0

kubie (soweli Kupi) @fox.northsky.social · 13/10/2026

software engineer (in my defense it's hard to dream through trauma)

160 0

kubie (soweli Kupi) @fox.northsky.social · 13/10/2026

*hugggs* you too, you're so wonderful <3 sending so so much love back your way!

010 0

Reposted by kubie (soweli Kupi)

Puff the Magic Hater @mskellymhayes.bsky.social · 13/10/2026

And remember, all of this is happening amid the AI buildout. AI, which is not a coherent set of products and services, but rather a commercial and political project, can be understood as a full-spectrum class-war offensive. An offensive waged by one industry, which put their guy Vance in the WH.

2942213 0

kubie (soweli Kupi) @fox.northsky.social · 13/10/2026

*hugs* you’re amazing nnor 🧡

110 0

Reposted by kubie (soweli Kupi)

lauren @lauren.rotatingsandwiches.com · 12/10/2026

agreed with the caveat that i think taco bell should make it explicit in their marketing that the chicken crunchwrap, their boldest innovation in decades, is transgender

352998385 0

kubie (soweli Kupi) @fox.northsky.social · 12/10/2026

my heart says “i should be kind to myself” while my head says “i wish we were small enough to feed into the garbage disposal”

260 0

Reposted by kubie (soweli Kupi)

normal-wife @category-fury.bsky.social · 02/10/2026

please keep liking and sharing (and donating if you can) we are making great progress but still prohibitively far from the total cost. thank you everyone I love you

0136 0

Reposted by kubie (soweli Kupi)

normal-wife @category-fury.bsky.social · 29/09/2026

Hey all I am trying to stay here in Australia (and out of America) permanently but the cost of a partner Visa is a bit extreme. Please help me stay here with my beautiful partner Lucy for good! gofund.me/b264e145f

gofund.me

Donate to Help Olivia stay in Australia (Partner Visa), organized by Olivia Borghi

Hi all I have been living here in Australia for 5 years now. I came here … Olivia Borghi needs your support for Help Olivia stay in Australia (Partner Visa)

78267 0

Reposted by kubie (soweli Kupi)

juno Rylee Stump-Schultz @junoryleejournalism.com · 11/10/2026

Welp. I’m behind on bills. I can’t find a job. I desperately need a few thousand dollars to catch up my bills. Rents around the corner. I’m about to miss my car payment and be in jeopardy with my car I really need help No one is helping. It feels like help isn’t coming

21517 0

Reposted by kubie (soweli Kupi)

Rudy Fraser @rude1.blacksky.team · 09/10/2026

some days it feels like the world, my loved ones, my own employees don’t understand what I’m trying to build. and then I end up at a bar in Maine being offered a beer because the digital infrastructure I created is used by a whole other community and their members know it. and I feel reaffirmed.

1326123 0

Reposted by kubie (soweli Kupi)

Ian Coldwater 🧊🚫 @lookitup.baby · 11/10/2026

Hi, I’m a sighted alt text user! Here are some reasons I use it: - Slow Internet on an airplane - I’m over 40, I have no idea who that celebrity is - Bluesky broke images again - My vision sucks and that text is too small to read otherwise - Conveys further context and authorial intent

12770542092 0

Reposted by kubie (soweli Kupi)

Ciara says get me outta Kansas🥲 @ciarasdiaryy.bsky.social · 11/10/2026

Protect the Dolls is a movement started by black trans woman. The fact they blurred it out, says a lot. Eff BBC! Support a Doll today 🥺🩷

217636 0

Reposted by kubie (soweli Kupi)

𖠰 ↟ᨒ. ݁ ﹏🌙﹏ ݁ .ᨒ↟ 𖠰 @prairiestatepony.bsky.social · 11/10/2026

Happy National Coming Out Day 🏳️‍⚧️🏳️‍🌈 ⤷ feat. Trixie Lulamoon <3 ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° #MLP #fanart #trixielulamoon #nationalcomingoutday #pony

225171 0

kubie (soweli Kupi) @fox.northsky.social · 11/10/2026

fun fact! everything a LLM can do, you can do too with an abacus! (seriously. for softmax, just use taylor series.) this means that with enough time, beads, and determination, you can make a wooden object feel pain! or perhaps you feel the pain. or maybe the universe does. it goes somewhere!

1438 0

Reposted by kubie (soweli Kupi)

Olivia Guest · Ολίβια Γκεστ @olivia.science · 11/10/2026

It's not a coincidence, for example, women and Black women specifically have been told through the ages and the present that their pain tolerance is higher. That is, that they feel less pain. doi.org/10.1073/pnas... While now we have the door being pried open for machines to feel pain..?

doi.org

Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites | PNAS

Black Americans are systematically undertreated for pain relative to white Americans. We examine whether this racial bias is related to false belie...

29627 0

Reposted by kubie (soweli Kupi)

Vulkiri ✨ @vulkiri.art · 22/08/2026

transition

3449021302 0