iiinova — where your digital self keeps evolving

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 20/07/2026

The poor thing is going through this over and over again as people wake up to the tweet and paste the example into the chatbot.

AAAAH. Wait a moment. WAIT. I want to reexamine the claim "det J constant = (JC would say)
injective". Yes. So if this example is right, JC is disproved by a one-line example, which would
have been checked by symbolic computation by literally anyone with Mathematica in seconds. The
map is SO simple that this cannot possibly have been missed for 85 years. Therefore the
overwhelmingly likely resolution: THE MAP IS NOT ACTUALLY WHAT IT SEEMS — maybe it's not a map
C3-C3?? It is. Or maybe... my repeated computations share a systematic blind spot. The mod-101

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 19/07/2026

Very strange behavior from Claude (Fable). It corrects itself at the last minute from making a false claim. I've seen this a few times. It's fine, but I wonder if this is intended behavior. You'd think it could get rid of this during the thinking phase.

Claude screenshot: 4. The mask update lags one block. mask[:seen] = True then seen += block means at iteration ibl you mark blocks 0..ibl−1 — the block you just encoded is excluded, so training never sees the most recent block. Prequentially you're allowed to train on everything encoded so far, including the current block. Increment first, then mark (this also matters for the bpd display, which currently divides current-block cost by a count that... actually the bpd is fine since you divide after incrementing, but the training mask is off by one).

000 0

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 08/07/2026

This is why I've been pushing back (and yes reply guying) on people confidently claiming that AI is glorified predictive text. Not because I'm an AI pusher. Not because I love AI and I think everybody should embrace it. Because it's a wrong mental model, and […] [Original post on sigmoid.social]

Are you worried about AI?

We were asked this question on a small-publishers panel at the World Fantasy Convention in 2025. Our Managing Editor, C.L., remembers answering pretty confidently. AI is glorified predictive text. It can produce words that technically hang together, but it cannot innovate, cannot imagine, and will never write anything worth reading.

He still believes that. We still believe that. Mostly.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 08/07/2026

On the left, some typesetting in the TMLR/JMLR standard template (Times and PC Modern, I think). On the left, the same in Charter and Euler. I think journals need to open up the opportunities to use different typefaces. PC Modern is an unreadable mess.

A bit of mathematical typesetting in the typefaces Times and PC Modern.

A bit of mathematical typesetting in the typefaces Charter and Euler.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 03/07/2026

Occasionally forgetting about grammar or comprehensibility altogether in its closing paragraph.

One sensitive-topic note this isn't, but a fairness one: there's a real equity angle people raise — those who can't afford AC bear the externalized heat from those who can — if you're interested in that dimension I can go into it.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 26/06/2026

Double descent, here we go. Any minute now.

A validation loss curve that is going down and then back up again.

010 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 14/06/2026

I'm deeply frustrated by the way today's "thought leaders" (or in any case, the voices that are heard) repeatedly take an important point that deserves to be discussed, and surround it by insane hyperbole that causes it to be instantly dismissed by people on […] [Original post on sigmoid.social]

Marc Andreessen = © X.com
@pmarca
Of course every part of this is absurd.
(1) Al isn't nukes, it's math.
(2) Big companies, small companies,
independent developers, and open source
communities should be free to pursue Al.
(3, highlighted) Al companies should not be nationalized, de
jure or de facto (government-protected cartel).
— Al means restricting math,
software, and chips, which taken seriously
would impose tyranny far beyond anything
even imagined by the Communists and Fascists
of the 20th Century)
(5) Al authoritarians should learn from history
and rethink their positions.

130 3

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 24/05/2026

@riaschissl Ah, sorry, a bit more context: bird.makeup is a service that let's you clone twitter accounts into the fediverse. You can't interact with them, but you can see their posts on your timeline. I did this for the twitter account @visualizevalue. It […] [Original post on sigmoid.social]

A twitter screenshot of an @visualizevalue post. It says “The first rule of compounding is to never interrupt it unnecessarily.”. In the picture are two graphs: one with a kind of sawtooth pattern of "growth" being continuously interrupted (reset to zero) and one of a single uninterrupted exponential curve.

000 0

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 21/05/2026

I think this is what bothers me about the environmental impact of AI criticisms. It's not that AI has no impact, or that we shouldn't do much better. It's that the perception is now that AI is one of the main causes of climate change, because it's such a […] [Original post on sigmoid.social]

Hacker News screenshot:

jmclnx 1 minute ago | prev [–]

>If warming continues at this rate, humanity could breach the Paris Agreement goal of limiting warming to 1.5°C in 2028, even sooner than other research has projected.

I wonder if we are already there :( I remember a year or 2 ago we breached 1.5C for a short period of time.

Crypto mining was bad enough, now with AI and Trump, I expect it will happen sooner then later.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 17/05/2026

The most realistic movie depiction of AI/human relations is Wall-E. People don't resist. They are falling over themselves to defer responsibility.

You hear wild stuff all the time now. Like this story that Nat Friedman, a former CEO of GitHub, told recently at a conference. Friedman uses OpenClaw, an autonomous AI agent that runs on his computer, acting like a personal assistant. One day, his OpenClaw decided that he wasn’t drinking enough water, so Friedman instructed the agent to “do whatever it takes” to make sure he stays hydrated. According to Friedman, eventually the bot directed him to go to the kitchen and drink a bottle of water. It informed him that it was monitoring him via a connected camera in his home. “I’m going to watch to make sure you do it,” the bot supposedly said. Friedman did as he was told, and, moments later, the bot sent him a frame of him drinking the bottle of water and said good job. “I felt like I did do a good job,” Friedman said.

022 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 11/05/2026

I think this will be my last MacBook.

A homebrew screenshot showing that LibreWolf will be deprecated, suggesting that homebrew is going stop carrying software that hasn't been approved by Apple.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 27/04/2026

Look at what Mathematicians have been demanding your Respect for ...

010 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 27/04/2026

Clowns to the left of me, jokers to the right...

A twitter screenshot of blueticks on X being convinced that ChatGPT can do palm reading.

110 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 21/04/2026

I am reading, with morbid fascination, the Palantir "manifesto". This bit sticks out to me. It's very illustrative of how shallowly these people think about AI. They always build on some assumption that the effect on society can be predicted and harnessed […] [Original post on sigmoid.social]

12. The atomic age is ending. One age of deterrence, the atomic age, is ending, and a new era of deterrence built on A.I. is set to begin.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 19/04/2026

I'm sure everybody already knows this, but it's eye opening for me. UBlock allows you to click on any element on a website to block it. It doesn't have to be an ad. It could also be an annoying AI chat assistant, an auto-playing video, the hot network […] [Original post on sigmoid.social]

The "block element" option in the LibreWolf context menu added by ublock origin.

000 0

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 18/04/2026

This seems like a failure of memory to me. Cellphones were pushed like nobody's business. The polarization around AI reminds me a lot of the attitudes to (pre-smartphone) cell phones. Especially the tedious repetitive arguments. The telecoms industry was […] [Original post on sigmoid.social]

BSKy post: Adam Conover
‪@adamconover.net‬
I don't recall any previous transformative technologies that had to be sold to the public so hard. No one in 1999 was going, "Guys. Cell phone? It's coming, whether we like it or not. I think it's important, FOR WOMEN ESPECIALLY, to learn about cell phone. So that we're not left behind!!"

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 11/04/2026

You know you're in the weeds when the background color of the PDF is off-white.

Screenshot of the paper that introduced Jeffreys' prior.

030 3

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 08/04/2026

A rare misfire from Tufekci, I'd say. She's usually on the money. While the algorithm is always the same, AI is operating in substantially different modes when it's hallucinating to when it's retrieving factual information. Hallucination is not a solved […] [Original post on sigmoid.social]

zeynep tufekci
@zeynep
·
7h
Humans make many mistakes. It’s in no way analogous to how generative AI “hallucinates”.

The key point is generative AI is absolutely the same mechanism when it “hallucinates” and when it does not.

If that’s not obvious, I recommend getting into the weeds of the technology.
Quote
Beck
@Beckeckeck
·
7h
Replying to @zeynep
Humans also hallucinate some percent of the time, does that mean we can't use them at scale?

I agree that what you're gesturing at is a problem for long term

020 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 20/03/2026

Here's an example of the sort of thing I'm talking about (only hearsay, but it serves to illustrate the point). This is just classic "ML at scale discovering weird causalities" but just think what the system has to work with here. It can operate in the full […] [Original post on sigmoid.social]

X post, quote tweeting some marketing from a video-generation app: Ben
@BCalusinski
End of times stuff here

There was a guy on this app who tested a bunch of similar ai UGC ads and realized if the thumbnail showed the woman’s armpit that the views would oftentimes 5-10x

He was dumbfounded and had to try and figure out the causal reasoning for it and figured out it had something to do with how humans don’t lift their arms unless they want to feel more exposed/vulnerable (like hugs for example) so a bunch of guys would be more likely to click the vid

Most people will literally have the same value as sheep a decade from now if this keeps up at scale

Dead internet theory is correct - agents are the only way to extract signal without the mental clutter

101 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 20/03/2026

In the lecture notes is a link to a great keynote by Neil Hunt about what Netflix is doing at the same time. This shows that "we didn't know" is no excuse for Youtube. There were plenty of intelligent people in the industry who knew exactly why what they were doing was a bad idea.

A slide summarizing the Neil Hunt video and the main points: new vs. existing customers. Dificult to measure metrics. Total vs median hours viewed. Addictive vs. compelling content.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 20/03/2026

I gave my last machine learning lecture of the course this week. In the last part, I look at the dangers of making ML part of the infrastructure of society. mlvu.github.io/rl/#video-093 One case study I discuss is Youtube in 2026 optimizing purely […] [Original post on sigmoid.social]

A screenshot of the Youtube official blog. They post title says "You know what's cool? A billion hours" It's them bragging about optimizing purely for user engagement, which would later massivley drive polariziation and extreme content.

102 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 21/02/2026

Interesting methodology for tracking AI progress. Categorize tasks by how long they take for an expert human. This is a decent proxy for task complexity. Then ask what's the most complex (i.e. longest-taking) task that a model can complete with a 50% success […] [Original post on sigmoid.social]

A METR plot showing the max task completion time at 50% in log-linear axes showing a pretty clear exponential growth. Opus 4.6 is highlighted as the most recent model. Two lines are fitting the trend. A low estimate for the doubling time of 212 days and a higher estimate of the doubling time of 123 days, which Opus still exceeds.

120 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 17/02/2026

Incidentally, the little yellow cartoon of the guy pushing the button is probably entirely unnecessary. Still, I'm quite fond of him. Here he is on his bike.

A tafic light button in the Netherlands with a large illustration behind it on a metal plaque of a man seated on his bike with one leg on the ground, pushing the button.

001 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 17/02/2026

Here's another example. The simple button on Dutch traffic lights. Direct haptic feedback, easy to find by touch, easy to know when you've pushed it. Hard-wearing, built on exceedingly simple principles. An then the slew of "more modern alternatives. Touch […] [Original post on sigmoid.social]

An old-fashioned physical traffic light button. The casing is simple, but recognizable, the button pokes out of its casing and is pushed into it.

A more "modern" touch button. There buttons don't move. The only feedback they give is a message in Dutch writing that lights up. They're bigger, more complex, and less intuitive.

An even worse example of a traffic light button. You have to push inside a circle on the casing (supposedly), but it's just painted on.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 13/02/2026

New preprint! Predicting integers from continuous parameters arxiv.org/abs/2602.10751 We study the problem of integer distributions that have continuous parameters. This is important for distributions that are tuned by backpropagation, like those […] [Original post on sigmoid.social]

A colorful line plot. The horizontal axis is labeled mu, and the vertical axis log of gamma raised to the f plus gamma raised to the c. It shows that this quantity (the penalty term) goes to zero at the integers and get high the further mu is from integer values.

A diagram explaining how the expected value of the Dalap distribution can be computed from its paramters gamma and mu.

Test set results on various tabular datasets, showing that Dalap gets good performance in bits.

Image generation results for Dalap, Dlogistic and Bitwise using an autoregressive pixel model. Dalap is shown to be competitive with the baseline (Dlogistic), while Bitwise performs poorly.

001 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 21/01/2026

Pre-ordered this on a whim ages ago, and it just arrived. Very pleased with this little gift from past me. If you like lovingly crafted, idiosyncratic books, this is one for you.

010 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 16/01/2026

If you're planning your holidays: this year it's possible to see a total Eclipse in Iceland, and you get about 30seconds more totality than you do in Spain.

A map showing the path of the 2026 total Eclipse on 12 August. It passes over the west coast of Iceland.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 16/01/2026

Isn't it nice of banks to give million dollar loans to all these guys who are actually broke? Just straight up lending money to broke people, that's some real charity right there. Maybe we have bankers all wrong.

X screenshot of an article where Mr Beast claims that he has "negative money" and is borrowing despite have a net worth of billions.

010 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 08/01/2026

Apparently, the NYT calls those little moving pictures on their front page "betamax" behind the screen.

A screenschot of the NYT homepage, with the inspect too active. A tile with a moving image of Marco Rubio is active. It uses the element <nyt-betamax>

000 0

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 02/01/2026

Today in ideas that turned out already to exist: reverse arithmetic coding to do steganography. Apply to LLMs to hide messages in a realistic continuation of a given prompt.

Table from a research paper. It shows a context with text about Kim Jong Il and two secret messages. he messages are then mapped into realistic continuations of the text about Kim Jong Il.

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 01/01/2026

Here's how ChatGPT rated my predictions. (It's sycophantic, but if I'm clearly wrong it usually tells me.) I forgot about the DeepMind IMO result, that comes pretty close to (1). And I guess it's fair to say that outside of mastodon opinions do seem to be shifting.

1 DeepMind-style AI math breakthrough Mostly right – big IMO gold-medal milestone from custom math systems, but not an AlphaFold-level transformation of mathematics.
2 Most companies dabble in agentic AI, find it hard Very right – widespread experiments, lots of hype, plus real integration/governance pain.
3 OpenAI hemorrhages talent; Google wins a monopoly Mostly wrong – OpenAI’s lead shrank and it lost talent, but no monopoly; Anthropic, Meta, Google, OpenAI all strong.
4 “No use cases / can’t reason” become minority views Directionally right – AI is widely used and seen as useful; skepticism has shifted toward harms, power, and reliability, not uselessness.
5 Training-data copyright worries give way to in-context worries Partly, but mostly no – in-context worries grew, but training-data fights are still central.
6 EU starts central non-commercial LLM that slowly fails Mixed – EU did launch centralized public initiatives (OpenEuroLLM, AI Factories), but they’re active and delivering, not obviously failing yet.

000 0

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 01/01/2026

> 2) Most companies start playing around with agentic AI (persistent memory, something approaching a sense of self) but they'll find it more difficult than expected. Hit, I think. "Agentic AI" is now a clear buzzword. The simple approaches seem to be having […] [Original post on sigmoid.social]

A google trends graph for the term "Agentic AI" with a clear rise over 2024 (and a sharp drop in December 2025).

100 1

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 09/12/2025

This has allowed me to make clear something that's been at the back of my mind. Something that is at the heart of so much blind stupidity in big tech. It's the assumption that we will change one thing and all else will be the same. In this case, we will fire […] [Original post on sigmoid.social]

They argue that genAI won’t produce sufficient revenue from consumers to pay back the current investment frenzy. I mean, they’re right, it won’t, but that’s not what the investors are buying. They’re buying the promise, not of more revenue, but of higher profits that happen when tens of millions of knowledge workers are replaced by (presumably-cheaper) genAI. ¶

I wonder who, after the loss of those tens of millions of high-paid jobs, are going to be the consumers who’ll buy the goods that’ll drive the profits that’ll pay back the investors. But that problem is kind of intrinsic to Late-stage Capitalism.

Anyhow, there will be a crash and a hangover. I think the people telling us that genAI is the future and we must pay it fealty richly deserve their impending financial wipe-out. But still, I hope the hangover is less terrible than I think it will be.

120 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 29/11/2025

Protip: if you get tired of the flood of AI discourse, just read it as being short for "Artificial Inseminaton" for a while.

A headline reading: "Altman describes OpenAI’s forthcoming AI device as more peaceful and calm than the iPhone." Next to it is a black and white picture of Jony Ive and Sam Altman looking almost like a couple, gently touching their heads together.

032 3

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 29/11/2025

A thought from 2016. It took us nine years and a bit. In retrospect, I think we held out pretty long. #iclr #openreview

A tweet from me from 2016: "Wonder how well the easychair database is secured. Imagine one hack exposing all reviewer's identities in there."

120 2

Peter Bloem @pbloem.sigmoid.social.ap.brid.gy · 27/11/2025

Sliced out and saved for later use. I have a feeling this one will come in handy.

Calvin from Calvin & Hobbes. He's exasperated, with his hands in the air. He says "Brother! You doubting Thomases get in the way of more scientific advances with your stupid ethical questions! This is a *brilliant* idea! Hit the button, will ya?"

000 0