@nielsa@mas.to

Niels Abildgaard @nielsa@mas.to · Aug 01, 2026

Replying to @glyph@mastodon.social

@glyph@mastodon.social I'm also reading it right now and it is very good: https://mas.to/@nielsa/116985893083797634

1

0

Niels Abildgaard @nielsa@mas.to · Jun 01, 2026

Replying to @abucci@buc.ci

@abucci 10/10 toot

0

Niels Abildgaard @nielsa@mas.to · Jun 01, 2026

Replying to @nielsa@mas.to

@baldur@toot.cafe Anyway, I don't think you're wrong, I guess I just have complex feelings about it 😅

1

0

Niels Abildgaard @nielsa@mas.to · Jun 01, 2026

Replying to @nielsa@mas.to

@baldur@toot.cafe I think there are way more interesting nuances to highlight than whether LLMs could be useful btw—it doesn't feel like a super significant point to debate right now. I'm more interested in strategies for actually making AI chat bot use less likely, pulling people out of delusion and addiction. That, too, can be a... let's say challenging... position to hold, when a lot of people are tightfully feeling upset with the adoption of this terrible technology. What is accountability and what is just vindictive? Do we provide paths for people who have become disilluaioned to actually join the community of people resisting AI use?

1

0

Niels Abildgaard @nielsa@mas.to · Jun 01, 2026

Replying to @baldur@toot.cafe

@baldur@toot.cafe Yeah, but at the same time that feels like ceding the field to the AI boosters, accepting the fight on their terms. It feels like when crypto bros made crypto mean -currency, not -graphy. Sure, it might be at best pedantic to insist on meaning in a time where people are busy picking sides... but when it comes to denying reality or not, I guess I'm a pedant. As a counter point, I see the exact point of LLMs as also useful tools being made by the folks at DAIR and e.g. in The AI Con (book). But made in such a way that they are still fully embraced by the people fighting AI: strongly denying and providing tools for fighting AI hype and LLM chat bot adoption. Those are the people I look up to in this.

0

2

0

Niels Abildgaard @nielsa@mas.to · May 15, 2026

Replying to @foolishowl@social.coop

@foolishowl@social.coop Haven't seen this. Got a link?

1

0

Niels Abildgaard @nielsa@mas.to · May 12, 2026

Replying to @pseudonym@mastodon.online

@pseudonym@mastodon.online @cammerman@mstdn.social If you ask the magic 8-ball about frequently truthfully described or discussed things, or things of similar structure with a clear mapping, it's more likely to produce a correct answer, but you have no idea what's frequently described and whether it's just randomly wrong. Oh, and if it is wrong, it is optimized to make the wrong answer look right in context. Good luck.

1

0

3

0

Niels Abildgaard @nielsa@mas.to · May 12, 2026

Replying to @xgranade@wandering.shop

@xgranade@wandering.shop @glyph@mastodon.social Aye. I am fully on board woth your take here. If we assume software devs who spend time posting about this or otherwise actively spreading the take, I agree 😁

0

Niels Abildgaard @nielsa@mas.to · May 12, 2026

Replying to @xgranade@wandering.shop

@xgranade@wandering.shop I fully agree with this 😁 I guess I'm just reading @glyph@mastodon.social 's post differently. I see a lot of people in tech landing on the position above as a way to not fully go with what, to them, feels like an overwhwlming pressure in the industry—and where they don't have the expertise to see the folly that. If they were pushing hard on that position, I agree it would be a problem. And it's a real separate problem that so mamy in tech do not have the expertise to discern, too. But different…

0

Niels Abildgaard @nielsa@mas.to · May 12, 2026

Replying to @xgranade@wandering.shop

@xgranade@wandering.shop @glyph@mastodon.social What does it mean, holding them to it in the future? I can't help but feel like that just sounds like wanting retribution. In my mind, getting out of the current situation and making "AI" use untenable in the future should be the goal... and retribution does not feel aligned with that. I certainly believe in accountability, but also that people can be misled and can change their minds. Am I missing something?

0

1

0

Niels Abildgaard @nielsa@mas.to · May 10, 2026

Replying to @emilymbender@dair-community.social

@emilymbender@dair-community.social @alex@dair-community.social Hell yeah, this is one of my least favorite "reasonable" takes, usually by non-programmers trying to do Serious Analysis.

0

Niels Abildgaard @nielsa@mas.to · May 08, 2026

Replying to @tante@tldr.nettime.org

@tante@tldr.nettime.org They're coming for established words again. Flow is the next crypto.

0

Niels Abildgaard @nielsa@mas.to · May 01, 2026

Replying to @nielsa@mas.to

@yosh Who cares if it's D or F tier? It's a significant degredation from a company that used to set the bar for the industry. Ultimately, the post hallucinates A Guy to debunk and is all splitting hairs while actually affirming the common sentiment, but framing is as the opposite.

0

Niels Abildgaard @nielsa@mas.to · May 01, 2026

Replying to @yosh@toot.yosh.is

@yosh That post is such a wrird take, as if (additive) uptime measurements aren't a well established standard? It's not even a takedown of the missing status page (which is based on how github used to report uptime, and helped popularize in the industry!) but the author's perception of people's perception of it? Zero nines undegredaded service is still bad. One nine uptime for most subservices is still bad.

0

1

0

Niels Abildgaard @nielsa@mas.to · Apr 28, 2026

ABCD: Always Be Collecting Data

0

Niels Abildgaard @nielsa@mas.to · Apr 24, 2026

Replying to @nielsa@mas.to

Those of us who have long been talking about the problems in software development are about to get a chance to convince a rapt audience, that has been burned by the existing dominant culture. Are we ready for that? Do we know what to say? How are we going to ghelp people become part of a movement for better software, for them and the world? I'm talking software development culture, inclusing for learning, sure, but also economic structures with better incentives. Now is the time to prep.

1

0

Niels Abildgaard @nielsa@mas.to · Apr 24, 2026

I cannot emphasize how much I believe we, as a software industry, need to be ready to catch people and offer alternatives once the AI hype bubble bursts and using LLMs for primary code generation becomes clearly untenable.

We had problems of incentives before. A lot of software was built bad or for bad purposes before. Silicon Valley was a huge influence on the whole industty before the AI hype, too.

The burst has the chance of disillusioning a lot of developers. Are we ready for that?

1

0

Niels Abildgaard @nielsa@mas.to · Apr 22, 2026

Replying to @baldur@toot.cafe

@baldur@toot.cafe I must have preread it: https://mas.to/@nielsa/116392028171325564 Or, similarly annoyed minds think alike...

1

0

Niels Abildgaard @nielsa@mas.to · Apr 22, 2026

Replying to @nielsa@mas.to

@timnitGebru@dair-community.social The reason I think of this whenever Dijkstra's Against NLP paper comes up/when LLM code generation is brought up: we already, as an industry/academy went through these motions before and found the idea of NLP flawed. Now we're doing it again, and trying to see how long supporters can keep this leaky blimp afloat before we inevitably reach the same conclusion. Again. I wish software engineers were taught more history of their field, I guess.

1

0

Niels Abildgaard @nielsa@mas.to · Apr 22, 2026

Replying to @timnitGebru@dair-community.social

@timnitGebru@dair-community.social Yessss! Another thread to pull on this: the terminology of "generations" of programming languages, which was first coined intuitively ("these languages seem like a leap forward in capabilities, but I can't explain why"), and was almoat instantly coopted by natural lamguage programming enthusiasts, who coined fourth and fifth generation languages as steps towards more natural language—which in turn led to the whole idea bevoming useless and largely abandoned.

1

0

Niels Abildgaard @nielsa@mas.to · Apr 19, 2026

Replying to @glyph@mastodon.social

@glyph I have definitely seen the effect in people I know.

0

1

0

Niels Abildgaard @nielsa@mas.to · Apr 18, 2026

Replying to @mjg59@nondeterministic.computer

@mjg59 I agree with this last statement 😁

0

Niels Abildgaard @nielsa@mas.to · Apr 17, 2026

Replying to @peter@thepit.social

@peter You're right, it's dark. But there are nice pockets out there where people care about the ability to do things well. The onslaught of "AI" inevitability discourse can be overwhleming, but it isn't all there is, I promise!!! Let me know if I can help in any way. I think you're right to want to stick to well-founded methods, and I think those will be valuable skills for you, if you stick with it.

0

Niels Abildgaard @nielsa@mas.to · Apr 12, 2026

Replying to @malteengeler@legal.social

@malteengeler Yeah, a lot of people are looking at the EUs moves towards digital soverignty as a solution to all our big tech woes, but it's like that cartoon about "more female dictators": changing the ownership of a rotten structure does not undo the structure. That said, a movement towards better structures can certainly take advantage of the mood du jour in the EU.

6

0

1

0

Niels Abildgaard @nielsa@mas.to · Apr 11, 2026

Replying to @yogthos@social.marxist.network

@yogthos It doesn't go far enough! I wanna get to that Old English-Old Norse convergence! Where's my original anglo-saxon(-jute) section??

1

0

Niels Abildgaard @nielsa@mas.to · Apr 11, 2026

Replying to @uncanny_kate@dice.camp

@uncanny_kate@dice.camp My brain read "libertarians" and got stuck for a full minute thinking "but they're the same thing" before I managed to read this right 😂

0

Niels Abildgaard @nielsa@mas.to · Apr 11, 2026

Replying to @libreoffice@fosstodon.org

@libreoffice@fosstodon.org

which might feel like an unfairness to some, even though it isn’t.

Pack it up boys, problems solved. You may have felt there was a problem, but there simply wasn't.

0

Niels Abildgaard @nielsa@mas.to · Apr 11, 2026

Replying to @baldur@toot.cafe

@baldur afaict it's even worse: requiring human experts after first triage, several exploration cases that are unlike real world scenarios etc

7

0

1

0

Niels Abildgaard @nielsa@mas.to · Apr 09, 2026

Replying to @mhoye@cosocial.ca

@mhoye Would love to boost if I could have an alt text on it? 😁

0

Niels Abildgaard @nielsa@mas.to · Apr 07, 2026

Replying to @pmakholm@social.data.coop

@pmakholm Ja, det kan jeg godt se. Måske er problemet at sociale medier prøver at være for mange ting på én gang for for mange folk. På den anden side: jeg prøvede at have en alt for at dele de to oplevelser op, så jeg kunne kuratere dem uafhængigt. (Nu kan jeg så ikke logge ind på @nielsa@norrebro.space længere såeeh)

0

Niels Abildgaard @nielsa@mas.to · Apr 06, 2026

Replying to @pmakholm@social.data.coop

@pmakholm Jeg er et helt andet sted, men har al respekt og forsåelse for at du er endt der! Jeg mener... ekkokammer -ammer -ammer -ammer -mer -mer -er - r

1

2

0

Niels Abildgaard @nielsa@mas.to · Apr 06, 2026

Replying to @mttaggart@infosec.exchange

@mttaggart I mean... the base assumption in the tri-system seems... unlikely to hold imo. Just from the abstract: "System 3 [AI use] can supplement or supplant internal processes, introducing novel cognitive pathways." That seems like a very generous interpretation of LLM use, that places it in a special category outside of other tool use, which is already covered by systems 1 and 2...

1

0

Niels Abildgaard @nielsa@mas.to · Apr 04, 2026

Replying to @rysiek@mstdn.social

@rysiek Thank you for looking into this! The site immediately struck me as conspiracy-brained when I first saw it going around, mostly because I found it difficult to quickly validate their claims. Thank you for going through this!

1

0

Niels Abildgaard @nielsa@mas.to · Apr 04, 2026

Replying to @Daojoan@mastodon.social

@Daojoan I have been thinking similar things, maybe framed a bit more broadly here: https://mas.to/@nielsa/116255178545615673 Would love your thoughts on that.

0

Niels Abildgaard @nielsa@mas.to · Apr 03, 2026

Replying to @GossiTheDog@cyberplace.social

@GossiTheDog afaik not a recent addition

1

0

Niels Abildgaard @nielsa@mas.to · Apr 02, 2026

Replying to @baldur@toot.cafe

@baldur Yes! And not just the code, but also what it implies about their API design, general application architecture, etc.

19

0

1

0

Niels Abildgaard @nielsa@mas.to · Mar 29, 2026

Replying to @aral@mastodon.ar.al

@aral@mastodon.ar.al deal

1

0

Niels Abildgaard @nielsa@mas.to · Mar 28, 2026

Replying to @aral@mastodon.ar.al

@aral@mastodon.ar.al Yeahh we don't fit those criteria (yet), but could maybe build something that fits if you can share the requirements

1

0

Niels Abildgaard @nielsa@mas.to · Mar 28, 2026

Replying to @aral@mastodon.ar.al

@aral@mastodon.ar.al I might be able to hook you up with some cheap VPSes hosted in Denmark if interested... we are working with a publicly owned data center that is trying to break into the private sector and care about digital/data soverignty. Let me know if relevant...

3

5

0

Niels Abildgaard @nielsa@mas.to · Mar 19, 2026

Replying to @cthos@mastodon.cthos.dev

@cthos @indutny I was thinking about the same thing, also vis-a-vis other replies here about how large and one-shot the PR is. Claude has gotten really good at producing code that *looks* right (that is, hides errors well), and trying to find them in 19k lines is just...

2

0

Niels Abildgaard @nielsa@mas.to · Apr 29, 2022

Replying to @Sandrockcstm@orbitalstation.one

@Sandrockcstm@orbitalstation.one i feel like at a UI level this could just be a big red warning? cc @Tusky@mastodon.social

1

0

Niels Abildgaard

Posts

Remote instance