Omar Antolín

I'm a mathematician at UNAM in Mexico City. I work in algebraic topology, homotopy theory and higher category theory, but am interested in all sorts of math.

I also enjoy computer programming as a hobby and am a big fan of the text editor Emacs.

Public Key

npub17mjld0zpmqv8uam9grchpt02kmgdwxlj7m3ede9r5ueefcstpcwqfjrup0

NIP-05 Address

oantolin@mathstodon-xyz.mostr.pub

Profile Code

nprofile1qqs0de0kh3qasxr7waj5puts4h4td5xhr0e0dcukuj36wvu5ug9su8qpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dspjr505

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub17mjld0zpmqv8uam9grchpt02kmgdwxlj7m3ede9r5ueefcstpcwqfjrup0

Show more details

Last Notes

2026-04-21 16:24:07 UTC

- reply

@nprofile…g0ws As the philosopher Neil Peart wrote:
Macros within macros in a spiral array,
a pattern so grand and complex,
time after time we loose site of the way,
our causes can't see their effects.

2026-04-04 21:02:37 UTC

- reply

@nprofile…l2h8 I think "more staff" is a very optimistic Freudian slip, and you actually meant "more stuff". :(

2026-03-27 14:41:18 UTC

Man, LLMs really drive home the lesson that people differ *greatly* in how much the care whether the statements they make are accurate.

2026-03-19 05:45:36 UTC

- reply

@nprofile…d0jr Here on Mastodon the video did not appear. It did in the RSS feed for your blog:
https://youtu.be/J-UUwG3L9IY?si=RUkK5loGg-s6kuXa

2026-02-21 22:49:11 UTC

- reply

@nprofile…4xpq @nprofile…g2n5 @nprofile…a6hr I think both things are true: politics in the US are heavily polarized between Democrats and Republicans, but also neither of them is leftist: the Democrats are centrists and the Republicans are far right. Occasional leftist politicians run as Democrats, not because that party is left-leaning, but because in that country there is only one other party and it is a far right one.

2026-02-12 22:00:40 UTC

I implemented org-babel support for 3 array languages I like a lot: Goal¹, BQN² and growler/k³. These languages are great for data munging so I find it very handy to use them in #OrgMode to process data stored in org tables. You can grab my implementations from my Emacs configuration at:
https://github.com/oantolin/emacs-config/blob/master/my-lisp/ob-goal.el
https://github.com/oantolin/emacs-config/blob/master/my-lisp/ob-bqn.el
https://github.com/oantolin/emacs-config/blob/master/my-lisp/ob-k.el
¹ https://codeberg.org/anaseto/goal
² https://mlochbaum.github.io/BQN/
³ https://codeberg.org/growler/k, forked from https://codeberg.org/ngn/k
#Emacs #ArrayProgramming

2026-02-10 20:02:58 UTC

- reply

@nprofile…tfwq hahaha, yes probably!

2026-02-10 16:10:52 UTC

- reply

@nprofile…tfwq Wow. Thanks for checking —it's something I didn't want to do on my phone.
Version 3.0 is so much faster, updating is low-hanging fruit.

2026-02-10 15:41:04 UTC

Why is LaTeX rendering on arXiv abstracts so slow? Are they still using MathJax 2.0?

2026-02-02 00:26:06 UTC

- reply

@nprofile…4hp5 @nprofile…t6k2 After reading what your wife thought, I took a look at some reviews and it gave me flashbacks of reading my student evaluations: one review sides with your wife on the need for editing, calling it "a tedious, bloated epic"; another review calls it "tightly constructed". When I taught undergrads in the US, I remember getting student evaluations, from the same class, one which said I was "the best math teacher I've ever had" and one which said "I don't know how the university lets him teach".

2026-01-31 19:39:14 UTC

- reply

@nprofile…pfuj I do 2x, except for certain accents which force me down to 1.5x. One thing I've noticed is that for complicated stuff, like a video of a math reseach seminar, listening at 2x forces me to pay attention whereas at 1x I often get distracted and have to rewind a bit.

2026-01-30 13:02:31 UTC

- reply

@nprofile…xuzd I won't use it all until I can submit to journals in Typst. Why learn a second document language similar to LaTeX when I don't need to? I learn programming languages for fun, not document markup languages (I guess Typst is both... maybe I should do Advent of Code in Typst —but definitely not mathematical documents until journals stop requiring LaTeX source.)

2026-01-29 05:31:13 UTC

- reply

@nprofile…xuzd I'd call it "conversion", in fact I think a lot of people already use that word when explaining what either "coercion" or "casting" mean.

2026-01-27 16:26:32 UTC

- reply

@nprofile…6pyy I found the author of that brilliant meme! It's Davide Castelvecchi, using a meme format devised by mathstodon don, @nprofile…tfwq! See this blog post for other nice fake math books: https://aperiodical.com/2022/05/didnt-graduate-texts-in-mathematics/

2026-01-27 14:37:45 UTC

- reply

@nprofile…6pyy
https://media.mathstodon.xyz/media_attachments/files/115/967/568/344/012/444/original/63e23b4fc6497186.png

2026-01-26 13:58:25 UTC

- reply

@nprofile…a3ey Did you the French Mathematical Society had decided not to attend the ICM? https://smf.emath.fr/actualites-smf/icm-2026-motion-du-ca

2026-01-25 22:37:39 UTC

- reply

@nprofile…xuzd Thanks for explaining that. It does sound very cool. I think the workflow you mentioned is something that I've avoided in git because the few times I've tried it, it seemed like a hassle.

2026-01-25 22:14:15 UTC

- reply

@nprofile…xuzd I imagine you were using git before. What makes Jujutsu so much more convenient?

2026-01-20 23:12:52 UTC

- reply

@nprofile…g0zc "Random find of the day". "Random"? I see what you did there!

2026-01-20 20:08:22 UTC

- reply

@nprofile…t6k2 Let's move on to something where we don't agree, then.

2026-01-14 15:53:15 UTC

I was so naive: I thought that once I passed my qualifying exams I'd never feel stressed by qualifying exams ever again! I failed to take into account I'd one day have graduate students, like them and empathize with them.

2026-01-10 06:23:25 UTC

- reply

@nprofile…a3ey I misunderstood, I thought the errata were missing, but only part of them are. Sorry for unnecessary comment.

2026-01-09 22:14:06 UTC

- reply

@nprofile…a3ey This doesn't answer your question, but one of the errata was probably about a gap in his proof of the Jordan curve theorem I found. We wrote a short corrigendum explaining the gap and two ways to fill it in: https://webhomes.maths.ed.ac.uk/~v1ranick/jordan/brouwer-cor-fun.pdf

2025-12-24 18:52:42 UTC

- reply

@nprofile…rrqe Not wanting reference documentation is complete unrelatable to me! My guess is that you either have a much better memory than I do, or are working with systems with bad reference documentation.

2025-12-22 16:11:19 UTC

- reply

@nprofile…w8t5 Not telling you the conclusion of their own work seems really weird to me! To a first approximation nobody cares what I do and nobody has read my papers (I'm hardly unique in this, of course), so I jump on any opportunity to tell people about my work. I would have expected others to feel the same way.

2025-12-15 02:15:27 UTC

- reply

@nprofile…ztfm I think after 6 months it is always an acceptable time to ask about a paper.

2025-11-20 16:20:26 UTC

- reply

@nprofile…rrqe @nprofile…h3ed You're right Jon that automorphisms of P_G^abs are of the form mu(_, u), but the map r(u) = mu(_,u) is not a homomorphism G --> Aut(P_G^abs), because r(u) o r(v) = r(vu). You want r(u) = mu(_,u^{-1}) instead.

2025-11-18 12:45:13 UTC

- reply

@nprofile…fhty I don't know, I kind of want the fields that use lambda to keep using it. It's a nice reminder that I'm leaving mainstream math and entering some adjacent land, like computer science or certain branches of logic,with their own language and customs.

2025-11-12 14:01:06 UTC

- reply

@nprofile…rrqe As a counterpointal anecdote, as a student I always wanted to know what was true even if the proof was too advanced for me at the time. A course with a large percentage of unprovable-in-the-course claims would be frustrating, no doubt, but I always enjoyed a sprinkling of trailers for future math.

2025-11-12 13:55:19 UTC

- reply

@nprofile…u6hy I work at UNAM (Universidad Nacional Autónoma de México), but not at UNAM (University of Namibia).

2025-11-07 04:54:09 UTC

- reply

@nprofile…fhty That's done with a nice Galois theory trick. Given an automorphism F of the field of complex numbers and a representation G --> GL(n,C) you get a new representation by applying F to all entries of the matrices. If you start with an irreducible representation the F-twisted one is also irreducible (because you could un-F-twist any decomposition). This means that automorphisms of C permute the rows of the character table, and this they fix the column sums. Therefore the column sums are rational numbers. But all entries of the table are algebraic integers; so the column sums are both rational numbers and algebraic integers, which makes them integers.

2025-11-06 06:18:57 UTC

Today I figured out a cute representation theory fact that must be well-known. The adjoint representation of a finite group G is a representation on the vector space C[G] but with the action given on the basis via g·x := gxg⁻¹. What's the decomposition of this representation into irreducibles? Well, the irreducible representation with character χ occurs with multiplicity given by the sum of the row of the character table corresponding to χ! I didn't even know before today that the row-sums of the character table were non-negative integers!

2025-10-25 20:44:07 UTC

- reply

@nprofile…rrqe That's not a reason against naming things after people, since the problem of having multiple inequivalent definitions also happens to term that are not named after people. In fact, I would guess that problem happens a little less with things named after people.

2025-10-23 21:01:34 UTC

- reply

@nprofile…ztfm It tricked me into spending a few minutes looking at it. I stopped browsing when the definition of the chain complex used didn't make sense. https://mathstodon.xyz/@oantolin/115418057236389832

2025-10-19 21:29:40 UTC

- reply

@nprofile…2k3a Why did you decide to add Helm support? What does the Helm-specific interface do that the completing-read version doesn't do under Helm?

2025-10-16 16:24:26 UTC

- reply

@nprofile…w8t5 For what it's worth, I don't find the nLab to be wrong often on topics I'm comfortable with. Wikipedia, of course is pretty good too, but there are many topics you can only find on nLab.

2025-09-10 11:27:56 UTC

- reply

@nprofile…yu7x Beautiful!

2025-09-08 04:30:29 UTC

- reply

@nprofile…59k6 I think I might hang out with too many constructivists on Mastodon. My first instinct was to correct you, to say you probably meant inhabited rather than non-empty, but... maybe not.

2025-09-07 02:52:03 UTC

- reply

@nprofile…6kk0 The first 3 of your 4 reasons proof assistants are interesting are some of the reasons I think mathematics (whether or not formalized) is interesting. Which I guess makes sense, but maybe I'm slightly disappointed your reasons aren't more specific to proof assistants.

2025-08-21 17:45:00 UTC

- reply

@nprofile…6kk0 I casually tried opening that link in Emacs's browser and learned Forester actually produces XML! And since you were complaining about the proposal to remove XSLT from the web platform a few days ago (from which I learned browsers come with XSLT!), my guess is that Forester is rendered by XSLT in my browser. That's pretty neat, but it means I need a browser with XSLT. Fortunately I can easily tell Emacs to always open your website in Firefox. 😛

2025-08-21 13:40:54 UTC

- reply

@nprofile…r9rs This reminded me of a story that Jon Bentley tells in More Programming Pearls (A very fun book!).
https://media.mathstodon.xyz/media_attachments/files/115/067/031/042/944/623/original/1b6d540cda6471cb.png
https://media.mathstodon.xyz/media_attachments/files/115/067/034/637/937/168/original/1eaacc4eec56a8c5.png

2025-08-06 19:49:28 UTC

- reply

@nprofile…6kk0 I did not know about metaweblog, it's good there's a standard.

2025-08-06 19:30:06 UTC

- reply

@nprofile…6kk0 The feed reader I use, Inoreader, has labels (but calls them folders, which is probably not a bad idea: if you call them labels people will ask you to implement folders :D) and does deduplication. I haven't tested the deduplication function of Inoreader for two different reasons: (1) it's only for paid users and I am happy with the free version, (2) I never see duplicate posts in different feeds (I do see duplicate posts within a single feed sometimes, though).
Being able to publish from your feed reader sounds intriguing. Where would those published feeds be hosted? Are you planning to add, I don't know, Wordpress integration or Micro.blog integration (both of which you mentioned)?

2025-08-01 01:43:26 UTC

- reply

@nprofile…pmdx I wonder if it'll be much smaller than usual because people are afraid to travel to the US. The Mathematical Congress of the Americas was held in Miami this month and some people told me that in some sessions close to half the speakers cancelled (I was one of the speaker who cancelled, so I had to ask a braver colleague who did go).

2025-07-31 14:36:51 UTC

- reply

@nprofile…fa4k Arguably this stick insect is both! Also, they've changed illusive to elusive on the website already, I only see illusive in the link preview here on Mastodon.

2025-07-30 17:33:40 UTC

- reply

@nprofile…m4ns @nprofile…e9rt I'd say that while these are techincally under C-c and follow the letter of the law of the question, they don't really count because C-c <letter> is meant for user key bindings. If you had some C-c C-<letter> that would be spicier.

2025-07-26 03:45:34 UTC

- reply

@nprofile…6kk0 It's funny that happens in functorial semantics. I never hear people say that in other categories: "a set is just an inclusion from the empty set", "a group is just a homomorphism from the trivial group", "a vector space is just a linear transformation from the zero space". Those things are all true, of course, but nobody finds them useful.

2025-07-25 01:54:29 UTC

- reply

@nprofile…pmdx Then I got nothing. :)

2025-07-22 17:20:26 UTC

- reply

@nprofile…v2av I don't understand: which field should have switched from using 1-category to what else? I have no guesses about which field you mean,but about what to switch to I would guess you mean 2-categories or (oo,1)-categories or (oo,oo)-categories.

2025-07-22 02:19:47 UTC

- reply

@nprofile…6kk0 Ouch.

2025-07-21 03:06:38 UTC

- reply

@nprofile…pmdx If people can think of an alternative section title that doesn't have formulas, rather than use that command you mentioned, I'd advise using the formula-less section title.

2025-07-21 02:33:44 UTC

You know how people advise you not to use mathematical formulas in section titles in LaTeX? Listen to them!
https://media.mathstodon.xyz/media_attachments/files/114/888/870/407/228/144/original/61e794b3e6e090c9.jpg

2025-07-20 17:36:03 UTC

- reply

@nprofile…6kk0 Do you know how it did it? Maybe a weight inversely proportional to publication frequency?

2025-07-20 12:37:46 UTC

- reply

@nprofile…6kk0 Oh, I see the comma and the truncated other feed name now.

2025-07-19 23:28:54 UTC

- reply

@nprofile…6kk0 I'm confused: how an I see that from the screenshot?

2025-07-19 03:59:09 UTC

- reply

@nprofile…pmdx That would work, but it's almost the same length as "(n-1)-connected pi_n-finite" which had the advantage of not even requiring a definition!

2025-07-19 03:01:50 UTC

- reply

@nprofile…pmdx I don't think that's too bad, you could certainly use that term, but it isn't what I would have guessed "virtually n-connected" meant: for spaces, I think of "virtually" as being related to finite covers, so I would have guessed that "virtually n-connected" meant the space is connected, pi_1 is finite and the universal cover is n-connected.

2025-07-17 03:00:43 UTC

It's a little odd to me that mathematicians use monads for their categories of algebras but functional programmers use them for their Kleisli categories.

2025-07-10 05:08:41 UTC

- reply

@nprofile…6cfc I didn't know that about the HoTT book but once I knew, I did immediately guess it must be Mike Shulman.

2025-07-09 18:47:03 UTC

- reply

@nprofile…hszg Well, ChatGPT is the poster boy for LLMs, so BC and AC would work, but to avoid confusion with COVID we could use L for LLM: BL and AL.

2025-07-09 16:23:07 UTC

I think I might edit my CV to add this next to the date I got my degree: "(5 years before the release of ChatGPT)".

2025-07-07 22:38:49 UTC

- reply

@nprofile…fa4k It bothers me a little bit that it has both "National" and "of Athens" in the name.

2025-07-07 18:23:28 UTC

ChatGPT kept telling people this software had a certain feature that it didn't have, so the developers decided to implement it. Normally *people* use ChatGPT to vibe code, this is the first instance I'm aware of of *ChatGPT* using people to vibe code!
https://www.holovaty.com/writing/chatgpt-fake-feature/

2025-07-06 03:45:07 UTC

- reply

@nprofile…t9cc For visitors I think the main risk is the entry, not the subsequent visit if you are allowed to enter.

2025-07-05 18:18:59 UTC

- reply

@nprofile…hszg @nprofile…6kk0 I mean, examples of that are a dime a dozen. It is very common to have something like "Theorem. The following are equivalent: ...", and then "Definition. A thing satisfying any of those equivalent is called a ...", and then to use the definitions interchangeably. It's very very common in most fields of mathematics.

2025-07-05 18:13:59 UTC

- reply

@nprofile…yms8 I'll let @nprofile…hszg answer in more logic-style language, but the topologist's answer is as follows: you are right that both definitions involve data, in particular, the correct definition involves the section and retraction and the homotopies proving they are such; the difference is what happens in each case when you consider the space of all possible data. For the correct definition the space of such data is either empty (when f is not an equivalence) or contractible (when f is an equivalence). The contractibility means the data is "unique up to homotopy" in a strong sense, which is why one says it isn't extra data at all. For the wrong definition the space of data still indicates whether you have an equivalence of not: the space is empty when f is not an equivalence and is non-empty when it is. However when the space is non-empty it is not always contractible and can even have many connected components (that is why Martin says there emight be several different g: g's in different connected components of the space are what count as "genuinely" different). If you truncated the space of data for the wrong definition to just be either empty or contractible, then you get another correct definition (this truncation is called propositional truncation in HoTT and in traditional homotopy theory would be called the (-1)-st Postnikov section).

2025-07-04 17:42:26 UTC

- reply

@nprofile…39xp Isn't it an LSP server? Ininagina that means no commands, you interact with it thought eglot or lsp-mode.

2025-07-02 05:30:24 UTC

- reply

@nprofile…yu7x Hahaha, I'm sorry, specially if I just cost you addition to heaven. 😅

2025-07-02 05:21:32 UTC

- reply

@nprofile…yu7x This gives new meaning and new higher stakes to the phrase "cinnamon challenge".

2025-07-01 21:37:38 UTC

- reply

@nprofile…hszg For what it's worth, I tried this prompt with Gemini 2.5 flash:
Please write a shell command that would tell me which files have the string ', refl' and also mention the string 'fiber' somewhere, recursively, considering only Agda files.
And it gave me this command:
grep -rl --include=\*.agda ', refl' | xargs grep -l 'fiber'
which I believe is correct for .agda files. Of course, it does not include .lagda files whose existence I learned about elsewhere in this thread (I assume it stands for Literate Agda). But I think I have slightly more patience than you for LLMs: I would use an LLM for this if I didn't remember the -l option to grep or the idea of piping to xargs; and I would consider that output satisfactory and would be willing to give it a pass on not including .lagda files.

2025-07-01 16:36:17 UTC

- reply

@nprofile…68z0 @nprofile…6kk0 I must have been really lucky: I don't think I ever read anything that made dependent types sound mysterious. I can recommend the HoTT book for example, it's very easy to read for mathematicians.

2025-06-30 21:13:12 UTC

- reply

@nprofile…hszg Fair enough.

2025-06-30 21:05:11 UTC

- reply

@nprofile…hszg @nprofile…v2av I think caring about the difference is an artifact of the affordances in type theory where it is convenient to always treat the identity types as the preferred notion of isomorphism. In more traditional mathematics, you set up the category of your objects to have the kind of isomorphism you want to consider and then only use that notion of isomorphism between them (there are of course exceptions when you want to consider more than one notion of isomorphism between your objects, which is usually dealt with by setting up various categories and functors between them).

2025-06-30 18:37:54 UTC

- reply

@nprofile…8atg Ah, so that's what it stands for! Obscure, frenCh progrAMming Language.

2025-06-26 21:48:38 UTC

- reply

@nprofile…hszg I does happen for me on the Mathstodon web page (Firefox on GNU/Linux), but my favorite Mastodon clients, the webclient phanpy.social and the Emacs package mastodon.el, don't preselect anything when replying.

2025-06-25 19:00:10 UTC

- reply

@nprofile…hszg How do you keep following the thread after you've been deleted? By manually going to the thread and checking every now and then to see if there's anything new? By hoping you catch new replies while browsing in your feed? Some people like being kept in the replies because the notifications help them stay in the loop.

2025-06-25 02:49:52 UTC

- reply

@nprofile…6cfc You're right of course a out BG in general, but sometimes BG does have the homotopy of a graph. It happens if and only if G is free. @nprofile…v2av

2025-06-25 02:33:36 UTC

- reply

@nprofile…6cfc I don't think homotopy theorists use the term "n-dimensional homotopy type" at all. There is the concept of "n-dimensional CW-complex" and therefore also the concept of "space with the homotopy type of an n-dimensional CW-complex". Maybe that's what @nprofile…v2av meant? As you point out Oscar, that is very different from being n-truncated (another term for n-type).

2025-06-24 19:19:28 UTC

A student sent me the evaluation forms I need to fill out for him partially filled out. Again. I had to tell him the government wants *my* opinion of his academic performance, not his own opinion. Again. 😐

2025-06-24 04:04:19 UTC

- reply

@nprofile…hszg Several apps crashing on startup sounds worse than an average beta, maybe the warning should be specific to this macos beta. @nprofile…6kk0

2025-06-23 19:58:40 UTC

- reply

@nprofile…6kk0 After the opening question I thought this was going to be a diatribe about logos. 😛

2025-06-21 20:56:21 UTC

- reply

@nprofile…hszg It's even true summand by summand (if you swap X and Y in the second one), right? I mean (X → 𝓤) = Σ Y : 𝓤 , (Y → X)

2025-06-17 18:50:00 UTC

When someone says: "There's this great dynamically-typed programming language with a useful and expressive type system, powerful macros and multiple-dispatch that solves the 'two language problem' (meaning that you can quickly and comfortably write a prototype in the language but then also write the fast production version in the same language), that has roots in academia but with uptake in industry, a great interactive coding experience in the REPL (including the ability to show you the assembly code for any function!) and [an] excellent compiler[s]" you don't know if they're talking about #CommonLisp or #JuliaLang until they choose either the plural or singular for the word "compiler"! 😛

2025-06-10 22:50:17 UTC

- reply

@nprofile…hszg @nprofile…mn9p I configure it to be an additional CTRL.

2025-06-10 12:53:45 UTC

- reply

@nprofile…6kk0 They're probably AI generated, it's cheaper than hiring humans.

2025-06-08 17:48:05 UTC

- reply

@nprofile…v2av Maybe the old-fashioned "set-valued functor" is more to your liking than "copresheaf"?

2025-06-06 21:00:52 UTC

- reply

@nprofile…saev @nprofile…6cfc There is this question from 2022, updated answers would be nice: https://mathoverflow.net/questions/430672/are-there-any-open-sourced-or-crowd-source-math-research-projects

2025-06-05 15:30:34 UTC

- reply

@nprofile…hr6z @nprofile…nalw @nprofile…uxsy What some people call "dead software", I think of as "stable"!

2025-06-04 15:21:59 UTC

- reply

@nprofile…39xp Journelly sounds great. I hope someday there's an Android version¹. You mentioned tags in your blog post and I was wondering whether Journelly supports org-mode tags or only hashtags (which I know it supports)?
————
¹ In my house only my son uses Apple products because my wife and I refuse to spend that ridiculous amount of money on gadgets for ourselves: my son's iPhone costs about twice the amount of my phone and laptop *combined*!

2025-06-04 10:36:50 UTC

- reply

@nprofile…39xp I like that the auto-generated alt text says it's ground meat.

2025-05-31 17:43:54 UTC

- reply

@nprofile…hszg That's exactly what has happened to me the few times I've tried that.

2025-05-27 21:04:40 UTC

- reply

@nprofile…hszg I'm sorry to hear that. For me this is, thankfully, an usually meeting-ful day. I usually don't have more than 3 or 4 meetings.

2025-05-27 15:16:24 UTC

What the hell?! I have 10 agenda items for today with a scheduled starting time? Am I an academic or an administrator? Is this what tenure is supposed to be like? 😅

2025-05-23 13:48:53 UTC

- reply

@nprofile…fa4k Damn autocorrect! Thanks for pointing out the typo.
Scotland? I somehow thought you had left the US for Singapore. At any rate, I'm glad you left the US.

2025-05-23 12:34:33 UTC

- reply

@nprofile…fa4k I think the article make a pretty convincing case that the brain drastic is already underway, it's funny they called it a "danger" in the title.

2025-05-07 02:56:46 UTC

- reply

@nprofile…59k6 Wow, what a lovely and unexpected answer!

2025-05-05 12:31:32 UTC

- reply

@nprofile…fa4k The author of that article, @nprofile…d6h6, was written plenty of sensible stuff on AI including a book, The Intelligence Illusion.

2025-05-04 23:17:27 UTC

- reply

@nprofile…hszg Right! This introduction's sin is honesty. @nprofile…rrr0

2025-05-04 20:43:08 UTC

- reply

@nprofile…hszg I can't speak for Dan, but in my case being aware there is no cause to feel embarrassment or anxiety doesn't really stop me from feeling them. @nprofile…59k6

2025-04-25 18:25:50 UTC

- reply

@nprofile…59k6 I understand why guessing at the underlying mechanism does not make a complaint stronger, but why do you think it makes it weaker?