@zanzi@mathstodon.xyz

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Jul 27, 2026

Replying to @johncarlosbaez@mathstodon.xyz

@johncarlosbaez@mathstodon.xyz @highergeometer@mathstodon.xyz @peterwoit@mathstodon.xyz @mc@mathstodon.xyz And even if there *was* such a thing as literal brain-enhancing steroids, using them to do math *still* wouldn't be the same as using steroids in sports, for the same reason that you mentioned.

2

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · May 01, 2026

Replying to @hegeliantaco@mathstodon.xyz

@hegeliantaco@mathstodon.xyz @julesh@mathstodon.xyz oh good catch xD

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · May 01, 2026

Replying to @hegeliantaco@mathstodon.xyz

@hegeliantaco@mathstodon.xyz @julesh@mathstodon.xyz this would be new, so please share it if you do it! We talked about this during the writing of the paper but didn't get around to it due to a lack of time and leftover space in the paper. But it would be for sure cool to see how cbv and cbn lambda calculi embed!

2

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 28, 2026

Replying to @boarders@mathstodon.xyz

@boarders@mathstodon.xyz Useless? It's the most important discipline around.

0

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 28, 2026

RE: @julesh@mathstodon.xyz

This is the conversation that mainstream media won't tell you about

3

8

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 25, 2026

Replying to @Andrev@types.pl

@Andrev @mc I agree that the essay goes beyond the scope of OP's post, but I'm more interested in the broader discussion than what OP is talking about, hence why I'm not even tagging OP in my replies.

Should we shame politicians or rich entitled people in their 50's? Maybe, but none of us have such people in our social circle, so any 'pressure' we try to put on them is mostly invisible to them. Meanwhile, I'm more reacting to the kind of sentiment linked in the forester above.

I am not interested in hearing any excuses for using the technology. If you feel compelled to say something like this to me, this is probably because deep down you know these technologies are harmful, both to you and to those around you, but you want to feel okay about how you use them.

As requested by the writer, I am also not tagging them, nor am I interested in engaging with them on this topic. So if their goal is to simply disengage from people like me, then they've succeeded, and that's a win win. But if the goal is to change someone's behavior, then the main behavior being changed is "I should talk to this person less", not anything to do with LLMs.

So tl/dr, shaming people is a trap, it feels good in the moment, but all it does is makes people less likely to listen.

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 19, 2026

Replying to @chrisamaphone@hci.social

@chrisamaphone @pigworker me and @julesh have a bunch of lambda and sequent calculi implemented for the category of containers, but would need to check which of them are public

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 19, 2026

Replying to @Andrev@types.pl

@Andrev @mc I saw this essay earlier, I will second that it's a good one to watch.

0

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 19, 2026

Replying to @liamoc@types.pl

@liamoc @mc I think this is mixing up two separate points. Students cheating on their home-work is bad, regardless of whether they do it by looking up the answer in a textbook, by using a GOFAI system like Wolfram Alpha, or pushing the magic answer-machine button. This is entirely *separate* from the point about shaming people in general for using LLMs across the board, which is... a bit cringe, and I don't think will help anyone. The other day I had someone try to shame me for even bringing up LLMs in a conversation without condemning them. I told them to take that sanctimonious preaching out of my mentions and not come back.

4

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 17, 2026

Replying to @pigworker@types.pl

@pigworker @cbaberle yep, same. there were rough times for a while, but i've always appreciated seeing you lead by example and call out the bs around you. and looking back, that's the stuff that stuck with me, not whatever arguments i was having with bigots at the time :D https://www.youtube.com/watch?v=EA-yqCCbTF4

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Apr 17, 2026

Replying to @pigworker@types.pl

@pigworker well, d'uh. when i was still early in my transition, it brought me a lot of comfort that the best type theorist that i know of was so vehemently supportive of trans people online. if i knew that she was trans, it would have blown my mind even more :D

10

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @zanzi@mathstodon.xyz

@hongminhee @jnkrtech You mention the idea that new crafts will replace the old, and I'm going to claim that new ways of *crafting code* will replace the old. We've climbed the rung of abstraction in languages, going from assembly to C to C++ to Python and to JS. LLMs are *not* a new rung on this ladder, because they replace how we *interact* with these languages, not the languages themselves. Instead, they reveal a need for new, even higher level languages, that will make JS and Python look like C in comparison. This way, an LLM could write a compiler not in a 100k LOC, but in 20k, or 10k, or however little is needed for a human to comfortably take over and oversee the maintenance of the project. I think *that* is going to be the future of software development in an LLM world, not a complete abdication of writing code to the models.

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @zanzi@mathstodon.xyz

@hongminhee @jnkrtech So while I very much relate to your dichotomy between craft and efficiency, I want to push back on your assumption that generating code through AI is necessarily more efficient. The Opus feat of rewriting the C compiler in Rust is impressive, and shows what you can do with an unlimited compute budget, but it misses the fact that the main way we engage with code day-to-day is not by starting new greenfield projects, but by maintaining existing code. And despite how good LLMs have gotten at one-shot generating applications from scratch, the more you start changing the requirements or adding new features, the more brittle the resulting code will be, and the bigger the attack surface for bugs will become. So the question to ask is, once an LLM writes a 100,000 LOC compiler, who will maintain it? If it's the LLM, well, then perhaps we will truly lose all connection to the code we produce But that seems too unreliable. But if it's a human, well, from talking to people who use these models at work, it seems that reviewing and maintaining LLM code is one of the most mentally taxing aspects of using these models. So whoever has to maintain this 100k LOC project is going to have a very bad time. My partner likes to joke that a good developer should be judged not by the number of lines of code written, but by the number of lines deleted. So like you point out, it's the *metric* of LOC written that's flawed, regardless of whether it's achieved by writing code manually or by using an LLM. But if the metric is no longer the amount of code written, then what should it be?

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @hongminhee@hollo.social

@hongminhee @jnkrtech Yes! These are fantastic, and exactly why I've asked for a *left-wing* analysis of this question. The right-wing/liberal pro-AI takes are dogshit, either blindly following the hype or idealising the corporations and billionaires who own this tech. I wanted to see an analysis that starts off from a place of caution, and looks at ways of reclaiming this tech, rather than accepts it as a social good apriori. And I'm not a Marxist, but I can appreciate their commitment to material analysis. I really like your idea of evolving new F/OSS licenses specifically for training (and your historical analysis of how the previous licenses evolved), I haven't considered that angle. I think it can be a good measure while the legislation/regulation is catching up. I'm also all-in on small, publicly owned models as an alternative to the frontier ones. I think there's a lot of food for thought here, and from your second post I can see you've already gotten feedback from a more political/activist angle. So I'm going to dig into the most conceptually interesting to me aspect, which is the capability of the tech itself. I'm a PL theorist and I work at an independent research lab that I co-founded with my friends. So I'm in the same boat as you when it comes to using these tools - I am free to use them to replace the menial parts of my job, and when it comes to code that I actually care about, that I write myself. But until recently, these tools couldn't even be used to write the kind of abstract code that I work on. Now... they can. Which is fascinating in its own way.

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @D3Reo@sigmoid.social

@D3Reo Yes, I think this gets to the heart of the issue, and this is also one of the reasons why I stayed away from this discussion so far. I think the framing itself is flawed. By arguing that these models are bad because they're not useful, we implicitly accept the framing of the tech CEOs for whom the sole metric by which we should judge this research is by how well it replaces human labor. But there's a lot of reasons why something could be worth studying. For instance, I used to love markov models as a kid, despite them not being particularly useful at modelling language at all. That being said, I don't think anyone owes it to any tech to be 'generous'. It's clear that it's useful for *some* tasks, but still bad at others. If it doesn't scratch someone's itch, it's fair for them to say so. And I personally find that it's much more interesting to figure out *why* it's good or bad at something, rather than tell someone that they're wrong about their own experiences with how they use these tools. The water argument *is* an issue, though. I think that's one of the key reasons to step away from frontier models and focus on developing smaller local models instead.

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @Andrev@types.pl

@Andrev Hype >:D I, too, have considered writing a manifesto. Happy to proof read it btw!

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @metarecursive@mathstodon.xyz

@metarecursive Yes, I agree that they have little to do with AGI, except perhaps as a single component of a larger architecture, much like the role that brocca's area plays for us. But the fact that they can offer a convincing simulacra of language *at all* is endlessly fascinating to me as a scientist. From the standpoint of philosophy of language, this gives us direct evidence that it's possible to separate language and reasoning. What's funny is that many people conjectured that you could have reasoning without language, but pre-LLMs I don't think anyone was seriously considering that you could have language without reasoning. And yet here we are. I haven't followed academic philosophy for a while, but I so wish to see it start tackling what this means for our conception of language. And I agree that as it stands, this technology is deeply tied to capital. That's why I think it's important to have these conversations now, before we get locked into this state of affairs for good.

4

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @Andrev@types.pl

@Andrev Well, you missed the after-office drinks with Jules and Orestis when this conversation started 😛 But yes, send me your takes on signal, I don't think this platform is actually conductive for a nuanced discussion.

0

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @aburka@hachyderm.io

@aburka You are in your right to think that. Now, since I'm so unserious, can you leave my mentions and not come back? >I don't see you elaborating on that Yeah, well, you're unpleasant to talk to. But you can feel free to follow along with the conversation, I'll elaborate once someone who is actually worth talking to shows up.

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @aburka@hachyderm.io

@aburka Well, I consider it both an insult, and a deeply unserious way to start off a conversation - by turning it into a pissing context of who is or isn't more left wing.

How would a position between "it's great" and "it's unfit for any use" not be a centrist position?

Simple, that's not a position that I advocated for.

I'm confused.

Good, it's okay to be confused. Perhaps next time, consider going on the offensive after clarifying what the person you're speaking to means to say, not before, yes?

3

1

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Replying to @aburka@hachyderm.io

@aburka Not everyone who disagrees with you is an enlightened centrist. If you can't imagine a radical use of a technology beyond what it was built for, then at least don't sully my mentions with your knee-jerk insults.

3

4

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 26, 2026

Where are the nuanced left-wing takes on modern AI and LLMs?

So much of the discourse around this tech is centered on rejecting it because of who currently owns it. But like all tech, it can be used for both oppression and liberation.

Who is focusing on the latter?

10

8

6

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 10, 2026

Replying to @mevenlennonbertrand@lipn.info

@mevenlennonbertrand very excited to read this!

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 06, 2025

Replying to @dimpase@mathstodon.xyz

@dimpase@mathstodon.xyz @dpiponi@mathstodon.xyz I'd say it should be an exercise for the author :D

0

2

0

Zanzi @ Monoidal Cafe @zanzi@mathstodon.xyz · Mar 03, 2025

Replying to @dpiponi@mathstodon.xyz

@dpiponi@mathstodon.xyz there's no way i'll even attempt to review this, but in all fairness, if such a proof does exist then I can completely believe that it could be done by an obscure fixpoint on metric spaces.

4

2

0

Zanzi @ Monoidal Cafe

Posts

Remote instance