Totally avoiding AI is getting harder, because Google and others for 20+ years didn’t move a finger to make search better. So free-tier access to the tools like GLM-5.2 provides helpful capability to search scientific publications on some topic. But putting one’s name on the output of generative LLM is another matter altogether.
As this was a response to my comment, I see it as insinuation I were an AI evangelist. I am not. I’m a user, which is by far not the same as evangelist.
Yes please. A bit more friendly and tolerant could also help
@Ronis_BR, I am sorry that this topic devolved in this way. I’m happy that you received this recognition and I think it is deserved.
I have used PrettyTables.jl myself and I truly appreciate the time and effort you have put into this package. I can only hope that my work could be as useful for others. Thank you for creating this package and please ignore anyone who would attack you for it. They are not important in the big picture.
Things are in flux, no doubt about that, but as someone who has been writing software from the age of 6 on my C64 and is still doing it today, I’m on your side brother 
Y’all, the level of respect for each other in this topic is far too low. No, it’s not ok to sling mud like this, and doing so will get your post flagged and hidden. Continuing to do so will lead to a timeout. It doesn’t matter the topic or the “side” — we must be more respectful of each other here.
23 Likes
Congratulation for the gift @Ronis_BR! This is presumably a drastically better vibe(or jazz?)-coder than all the other free/low-cost alternative. I would be curious to try it out.
I’m not sure it’s a good idea to step in the pit, but since you asked: no, the “not working” threads
are not disproportionate. The usage increases and so are the bug reports. This is completely normal, and as a FOSS community we are transparent on everything, including bug reports.
It’s the promotion. My advise is to slow down on the promotion of your package (but not on the development, this is super useful for the Julia community!). Particularly, self-promotion on threads started by others, when the initial question was not related to AI. It can engender the opposite effect of what you want: people will run away from it.
Many of us in STEM genuinely want to keep an open mind but experience a deep exhaustion of all the AI promotion, too often presented as a panacea without any nuances. And we close ourself as a self-defense mechanism (BTW, this is what scientists instinctively do in front of too-good-to-be-true outcomes). The frequent self-promotion piles up on the noise and we conflate it in the basket of “one more guy who proselytize AI”. When (I assume) you just want people to try your package because you sincerely think it is useful for us.
TL;DR keep up the good work and keep us informed in the package announcement section!
Congratulation on the recognition, and thank you for all your packages:
No. Strictly not true. You can now run Kimi K3 on your CPU with 8 GB of RAM (yes slowly…). I believe you can maintain without AI, or use local (or cloud) AI, or just ignore such packages. It’s a package problem at best, ecosystem is a mix of those and not done by AI. Either way a free gift.
AI is limited by electricity (and HBM DRAM), and electricity/grid by electrical transformers, so we’ll not get more data centers fast. The transfomers take 2 years to build, each is custom and weigh many tons. I would worry more about AI pushing electricity (and [DDR] DRAM) prices up, and inflation, than water use.
Global water use is a problem (at least local problem in many areas), but don’t blame AI, and it’s best to distinguish consumptive and non-consumptive use. Water use for AI in data centers is e.g. dwarfed by water use for golf courses. It’s use is “peanuts”, also dwarfed by literal growing of just peanuts (not all food), and use for peanuts dwarfed by growing corn or alalpha (separately, see graphs in following article).
On the national, local, and personal level, AI is barely using any water, and unless it grows 50 times faster than forecasts predict, this won’t change.
https://www.sciencedirect.com/science/article/pii/S0043135426005488
Technical solutions exist but deployment remains limited by policy gaps.
No, can you name one?! [EDIT: They bought millions of used books (not illegal); at least their service provider (Google?) did the destroying (not illegal to destroy what you bought, only what followed and training controversial) and scanning (they claim fair use)" https://s3.documentcloud.org/documents/25982181/authors-v-anthropic-ruling.pdf “Anthropic spent many millions of dollars to purchase millions of print books, often in used condition. Then, its service providers stripped the books from their bindings, cut their pages to size, and scanned the books into digital form — discarding the paper originals. Each print book resulted in a PDF copy containing images of the scanned pages with machine-readable text”] You mean it “read” the books (was trained on), at best a copyright violation, I think still undecided by a court if copyright then applies.
EDIT2 (from the ruling):
This order grants summary judgment for Anthropic that the training use was a fair use.
And, it grants that the print-to-digital format change was a fair use for a different reason. But it
denies summary judgment for Anthropic that the pirated library copies must be treated as
training copies.
From Ars Technica (thanks @ufechner7):
The first-sale doctrine offered a workaround: Once you buy a physical book, you can do what you want with that copy—including destroy it. That meant buying physical books offered a legal workaround.
Well, they bought them, scanned them, and threw them away. OK, if you call that destroying, that is a subjective matter.
Well deserved!
The purpose of a system is what it does.
Is the only reasonable and remotely insightful thing ever spoken by a management consultant.
The problem isn’t with AI, or its use. The problem is the system that socialises risk and privatises profits. It’s the system that convinces you to freely contribute your time for “the greater good” when that good is concentrated in the already rich and powerful.
It’s the system that alienates people from themselves, their communities, their world, and their labour so much that we become husks trying to hustle what little scraps are left for us. A world where top rate talent can’t make ends meet because they are neurodivergent, or they fell on hard times, or went through trauma without mental health resources, or they lost the post-code lottery.
Still that talent gives away their time, their life, their ideas, and their work for their communities.
AI comes with a lot of ethically reprehensible, dubious, and flat out horrifying baggage. But the people in this community are not using it to create AI misinformation. They are not using it to spread hate, or division. They’re using it to give humanity something more than what they can give on their own.
Yes, it uses a lot of energy and water and that fucking sucks. In the first three decades of my life I saw my home town aridify to a scary degree. It went from a tropical savannah/subtropical wetland to a hot desert. The lakes dried out, trees were felled, the wildlife died. That happened before AI.
AI is another drain, yes, but it pales in comparison to the fact that there are nearly as many cars in the developed world as there are people there. Hegemonic society keeps drilling for oil, keeps making it necessary for you to buy a car so you can buy groceries, it freeloads off the good will of FOSS and barely gives back. It incentivises and rewards grifters, it has convinced us to blame other victims of the system for our problems or the fact that the god damn Danube is bloody drying out and multiple countries are in literal flames right now.
The purpose of a system is what it does. And the system has convinced us that it is our individual choices that have got us in this mess, and not the “benefactors”, the billionaire “philanthropists”.
Remember this before attacking other people who—like you—want what is best for the communities and are trying their hardest despite being spat in the face by the hegemon pretending to have their backs and being very “grateful” for their contributions, while profiting from them.
This is not to say AI is completely without fault, i’m always on two minds when using it. But we must keep in mind the problem has been around since much before it. And other areas contribute much more to the sad state of the world, and its sadder trajectory.
As Yanis Varoufakis put it, “we want your AI, we just want to expropriate it!”
What about: do not annoy people about what their are using to develop packages in their free time, giving work back to the community for free? Unless, of course, you are sponsoring the project.
Funny enough, this is backwards! The anti-AI people is always hijacking threads for the sake of opinion without pointing to this date a SINGLE TRUE problem in using AI in Julia community.
Just to describe some facts: the behavior of people like @Mikhail_Kagalenko and @kellertuer just made Kahli lock / delete his packages (Tachikoma.jl, Kaimon.jl, etc.). Julia lost something truly amazing here today. It kinds remember me nvim-treesiter. When this community became so toxic?
Well, at the end, Anthropic gave me 1200 USD (which is a huge amount of money in Brazil for open source software development) and I am really improving PrettyTables.jl and SatelliteToolbox.jl. Besides that, I had received nothing from any other foundation. I also have never received anything from you, not even a single PR.
Thank you very much @Tamas_Papp! I am really glad that PrettyTables.jl is being useful for you. Btw, my next feature will be org-mode backend that you asked me soooo much time ago 
Sorry for the delay.
Fable 5 is at another level. It found ways to optimized time to print the first table that I would never be able myself. No other model that I tried (including GPT 5.6 Sol) found this optimization.
It seems to have been made private, the forks still exist. That is a bit sad, but unfortunately not uncommon in open source. Hopefully they come back soon.
In these discussions I’m always on two minds too. This is the only comment which almost resonates with me in its fullness! I admire that you tried to reconcile people with this idea.
Sorry to be obscure: what I meant is that your use of AI produced a positive benefit that the intolerant would condemn because it was somehow ritually unclean. I think it’s great that you were able to genuinely improve a valuable contribution by using AI as it should be, as a force multiplier, not a substitute. Apologies for the misleading comment–the snark was directed at the Luddites.
I agree it’s highly problematic that many topics that touch on AI usage have been diverted into turf wars about AI itself. We’ve been trying to moderate such things — both to be on-topic and respectful — but it’s been challenging and too slow. By the time flags hide a post, the train has already derailed and a dozen traincars have already joined the flames and brought even more gas to the fire. At that point, nobody involved is happy about any moderation because they’re “just” following the prior car.
I’m seeking input from others on how we might better prevent such train fires in the future.
In the meantime, let’s please avoid “directing snark” at anyone at all.
@Ronis_BR That’s awesome! According to the package ecosystem knowledge graph that I created (with the help of Claude 
), PrettyTables.jl is in the top 50 (#46) most depended on packages in the registry (out of 14,000+ packages). It has 227 direct dependents, and 2,888 transitive dependents!
as a former discord moderator in a big discord server, the way to prevent such arguments is to build trust in the community and especially between the staff team and the community, so i would suggest focusing more on building trust with the community, that is from my experience and i am not sure it will work here, but maybe its worth a shot!
I’m on two minds about using it myself (i still do) but i will never blame people for using tools, especially when they give away what they make with those tools for free or a reasonable exchange. I will absolutely blame a system that incentivises and rewards those who use shady tactics to create the tools and puff out their chests when they offer tokenistic gestures like gifting 1200 USD to individuals after they gave their lives to a project for free for years in darkness and solitude. And puff their chests out as if they’re good people.
As someone who is inclined to argue that current mass-use LLMs are ontologically evil technology with even worse embedding and material impacts on current society, I agree with the sentiment that the prevalence of AI discussions here on the Discourse make me less inclined to browse it, categorically.
That is definitely a challenge. However, I think there’s some intrinsic challenge or contradiction in trying to steer towards neutrality regarding a technology that is incredibly political. It is also a technology that many materialist and politically literate technical people are inclined, as I mentioned above, to view as ontologically problematic and even more harmful in its application.
I would expect that many people will simply not agree that discussion of LLMs (at least ones in the neighborhood of the current art) should occur as if the broader context does not exist: hence, your observations.
I agree, the post definitely stands out. However, with all due respect, I am afraid it is not mutually exclusive and collectively exhaustive. Nevertheless, it contributes a valuable voice to the broader discussion. The parallels to Post-Keynesian and Modern Monetary Theories sound almost like a prioritization of central activities that shape markets to achieve societal goals, rather than allowing free markets to determine economic outcomes. However, I am not entirely sure if I have deciphered all the paragraphs correctly, particularly regarding the perspective on capital controls.
Anyway, from my point of view, while I don’t think this is intelligence like ours, it seems that we are witnessing a structural change and in the interconnected world this change is rather not going to go away. For me, the question is: how can we utilize / leverage this innovation to drive the growth of the Julia Programming Language?
I use them too quite a bit (I’d like to use more the chinese models in the future since I like the way they are handling this more).
From a purely technical point of view, LLMs are great, they are clearly very useful. We have confirmations almost every day of this in different forms e.g. their ability to resolve some ancient math conjectures. My main concerns are though much more political/societal. I could see them as quite useful tools, and at the same time, questioning non-technical aspects of the technology as e.g. the impacts on education or how it is developed/trained.
Indeed I very dislike the ad-hominem attacks to @kahliburke, who, as many of us, is enthusiastic about that now some things can be done better and faster (hope more the first than the second actually).
On the 1200 USD, this is a great gift, but seems as poisoned as the fact that I can pay 20€ a month and have an amount of tokens that if I paid through the API would be 1000€. This seems a boiling frog strategy (fortunately we have deepseek and other labs), or maybe it is just that API prices from some providers are so so inflated, but still seems the same actually even if so. I’m very happy though that they gave them to him, I also applied to a similar initiative but they didn’t give me