Clojurians Log v2

Of course there were. :)

You are right, yes.

This was just like an AI conversation 😄

@p-himik was right to push back 😂

Yeah, right now there's a window where technologists can publicly criticize this new tech And not so much AI, but rather "AI minus other tech" — that subset of AI uniquely unlike other tech Also, criticism usually focuses on the tech as much as possible, not inquiring much into features of the society wielding it. So, we don't ask "Under what conditions would this tech be more utopian than dystopian?"

I think one thing its telling is how electricity and internet were less controversial than AI

@tjg, have you read Chomsky's letters to Epstein? I have zero problems admitting that I do not care at all about Chomsky's opinions (never cared too much either) after the letters. Homo sum, humani nihil a me alienum puto.

You have valid points and I agree more or less with you: almost everything is crap.

So "educated" people dislike the current LLM trend and "people that do not know any better" do? Do you realise where that viewpoint might take us to?

All I know is that LLM madness led by VCs is leading us to financial, cultural, ecological and cognitive disaster.

And yes, I do believe that the most capable, the most creative people dislike LLMs.

And this too:

I am not a fan of generalising this way. People can have miriads of reasons for using or not using LLMs. They can also be well educated or not, experienced engineers or not. And there are far worse ecological disasters in the world than llm use. And perhaps more importantly this framing stops any form of discussion if we are not careful.

It's just an observation. I don't understand how susceptible people are when you criticize LLMs. It has cultish vibes. Guillermo del Toro: https://www.instagram.com/reels/DTK47NyDQYk/

Regarding ecology, have you read how much electricity and water these datacenters need?

I am not denying the potential disastrous effects on the environment or on peoples behavior. But it is too easy to be judgemental about it. Throwing random quotes doesn't really do the debate justice. As a human being we leave a footprint, no matter what we do or not do. I have not eaten meat for more than 30 years. I stopped using airplanes, try to live with as much from our garden as we can, and I happen to have a PhD in Computer Science. Studied and used AI. I think I have a reasonable education and have spent most of my professional life (I'm closing in on 60)d developing and working with software and software teams.I am not a fanboy of LLM although I personally see interesting uses. I think it is good to discuss the potential effects of LLM use, but the world isn't going to pieces because of just that. There are far more urgent and potentially dangerous things out there. We are responsible and we should also take responsibility for using this tech or not.

Just to be clear I do not see myself as a Sr engineer at all. There are way more smarter and experienced devs out there. I'm more of a casual developer as I also have a lot of other responsibilities

Quote from the clip with Guillermo del Toro: > The greatest crimes to civilization have been done in the name of convenience That resonnates with me. You see it in the small and in the large, in software ("simple vs easy"), in schools ("it's just easier to use ipads than pen and paper" - easier for the admins that is), everywhere. If it's out there, then it'll be used. Nobody cancelled Musk for being a white supremacist - here in Norway Tesla sales went up instead of down - it's just so damn convenient to own and use those Teslas. And nobody is giving a damn about the effects of using LLMs as their daily driver. At least, that's how I percieve the world at the moment. Can something be done about it? Sure! It can be regulated. Give penalties/taxes for ecological effects. But there's too much money to be made, and people doesn't seem to care about the future as long as their life is undisturbed today. /end rant

Those are good observations Gunnar. I am inclined to believe that in the end we (humanity) need to decide what is ok and what not. Governments or politicians won't act until we tell them to act. I also drive an electric car (not Tesla) and it feels much more comfortable for me than a car on fossile fuel. Does that mean it has less of a footprint? I don't know.

I generalize in order to have meaningful conversations, to reflect on stuff. To have a voice. Gen AI is the perfect excuse to destroy jobs, to make us servants to monopolies, to make us mediocre and, according to my gut feeling, the people behind it (Nick Bostrom and the like) have a sordid hidden agenda. Who is winning with LLMs? Who is creating awesome things with LLMs? I have no problems apologizing or changing opinions when proven wrong.

I don't pretend I know how to handle these discussions but in general I feel it is better to concentrate on discussing the tech and it's consequences and less about the people using it to whatever end

I take note. I've read many "programming languages do not matter" in this slack and no one has raised any issue with those comments. What a world.

I have no idea why anyone would say or believe that at this time.

> no one has raised any issue with those comments One such comment was a catalyst for the creation of this very channel. So issues have been raised, plentifully.

I was not aware about that. Thanks.

> "programming languages do not matter" I don't believe that phrase is valid now, but I don't think it is a crazy one, and could be valid soon. Programming languages are just tools for humans, our current way to automate things with computers, but I don't think it is crazy to think it could be replaced by some other way of telling computers what to do.

Feasibility is always a concern, like with any kind of tools, even outside of IT. You can build a house with just a flat screwdriver and pliers. But it's inconceivably harder and would take an incomparably large amount of time compared to using proper tools for the job.

I saw you wield metal the other day p-himmick. Talking about the proper tools!😎

@jpmonettas, it will be very pleasant for the Clojure core team to read this kind of comment in their Slack.

@p-himik But soon computer languages could became building a house with pliers. You can, but inefficiently. As I see it, computer languages trade off wasting energy (being sub optimal on the final set of instructions) by helping humans to reason about those instructions, so we can build better things. But if you can get other ways of expressing what you need to build and how to test it into crazy low level instructions you could potentially get rid of the entire stack (langs, compilers, etc).

How soon, @jpmonettas?

@asier.galdos I don't see it as binary, all or nothing but a transition.

Wonderful.

> But soon computer languages could became building a house with pliers. I'm not comparing "humans building stuff" to "robots building stuff". I'm comparing two hypothetical things that can be used by anyone or anything. An AI would also have to use something to make a von Neumann-architecture PC do what it needs it to do. > But if you can get other ways of expressing what you need to build and how to test it into crazy low level instructions you could potentially get rid of the entire stack (langs, compilers, etc). Would it pass the "compiler goose test"? If it quacks like one, smells like one, and looks like one, it's probably a compiler. :)

I want to be clear that I'm not here on the skeptics to defend the use of AI, I think there are a trillion problems around it. But I think arguments are important for all of us learning about the thing

I am pretty sure new languages with new properties will appear, but I cannot imagine (myself) yet another way of dealing with computers. Perhaps you can plug your brain in at some point but there is still the problem of understanding a problem and formulating a sound solution.

If some hypothetical AI receives an instruction in plain English or whatnot and spits out a binary that can be run on a target PC, that AI has just become a compiler. With English being the programming language - ambiguous, underspecified, shitty language, but a language nonetheless.

And in that sense, language do still matter, and will always matter.

> Would it pass the "compiler goose test"? If it quacks like one, smells like one, and looks like one, it's probably a compiler. depends on your definition of a compiler. Compilers on my head are a formal language -> other formal language. LLMs and other tools like it could be more like give me diagrams, images, text in any language, pseudocode, and whatever you have and I can give you a well optimized set of instructions for X processor that passes whatever we defined as tests. So you can call that a compiler or not.

With English being the programming language - ambiguous, underspecified, shitty language, but a language nonethelessbut why it has to be limited to english? I mean you could do english, images, pseudo-code, python, diagrams, a video of someone demoing a physical interaction with a mock device, etc

> but why it has to be limited to english? It doesn't, that's what I meant by "or whatnot".

> depends on your definition of a compiler In that case, the whole discussion is about semantics. :) X matters as long as X can be defined as a wide enough thing. X doesn't matter in the context of a narrow definition of X when that definition ceases to be useful. Maybe the latter will happen to the currently existing programming languages, maybe not - unknowable at this point.

My argument was that being able to specify your idea using a bunch of different media, refining conversations (including some formal languages) , could eventually replace programming languages, not that they will be completely useless, but more in the vein of woodworking hand tools are still useful but their impact in the world furniture building is very low.

@asier.galdos Yep I've invaded his privacy by reading those private letters, from the perspective of someone interested in the toll that activism takes on people's lives. And I actually like the guy more now! I could explain why, but this is #C0B7Q4XRM1U; gossip about Chomsky (or Sussman, Walmart, Buddha or Hitler...) doesn't help us find truth on this topic If a main skeptical technique is https://en.wikipedia.org/wiki/Pyrrhonism, then focusing on celebrities/personalities is a hasty reason to judge critiques of technology This thread's premise is that "There were no electricity or internet skeptics in the past." But I think we're blind to the history of tech-skepticism except AI-skepticism — and even then, we probably only focus on narrow politically correct slices of it

Interesting. I will read the link tonight. And feel free to share gossips. I am intrigued now.

Regarding Chomsky, I think the best source is Bev Stohl, his teammate who saw every email he sent. She wrote: https://bevstohl.substack.com/p/im-no-longer-waiting-for-the-storm Regarding evaluating someone's morality, what's our evaluation procedure? Mine's based on how much suffering one helps reduce. My analysis is too long for this slack, so I https://gist.github.com/caliadero/1ff1c349885cea653368cc9d2390ab69. I had limited time to edit it, so please read charitably, if you do read it

@tjg, just read https://gist.github.com/caliadero/1ff1c349885cea653368cc9d2390ab69. Thank you for writing it.

@tjg after the Epstein files my views on Chomsky changed forever. Thanks to the internet access to knowledge was already available to everyone. AI gives you everything pre-digested. That’s what people like. People don’t like to think too much about things, especially about things outside their area of ​​knowledge. The same goes with newspapers, it’s a completely different experience to read something you are familiar with than about something you know nothing about. When I say “normies” (I should probably use another word) I mean conventional people, trend followers. My memory is that the early internet was more organic, more nerdy, more genuine, and conventional people joined the party later. Now that our industry is trendy (and lucrative), conventional people are the ones pushing for AI.

(Chomsky had been a highly controversial figure well before the files were released. I would triple check everything he says, with a potential exception in linguistics.)

Focusing on personalities (not points) is part of the enshittification of the social sphere — ChatGPT is more reasonable here than humans. So's the sciences (including linguistics), where it ideally doesn't matter much if Buddha or Hitler proposed a theory Walmart was notorious for decimating towns and livelihoods, but here in the Clojure community, we use their opensource libs and https://news.ycombinator.com/item?id=19728502 how Clojure https://www.cognitect.com/blog/2015/6/30/walmart-runs-clojure-at-scale. And I think that's reasonable, though interestingly there's no #tech-skeptics channel to discuss how our work impacts society Yesterday in #C068E9L5M2Q, I cited Sussman, whose work on Scheme probably influenced Clojure/script. But since I was making a technical point, no one cared about Sussman's https://en.wikipedia.org/wiki/Gerald_Jay_Sussman#Academic_work But here, we don't discuss Chomsky on-topic points, but rather someone he associated with when he was nearly 90. There's a suggestion to triple-check his points — which we should be doing anyway like code review — but I think that's rhetorical. So let's check them: • Internet was funded by the public and handed to private sector: analyzed by Mazzucato's https://en.wikipedia.org/wiki/The_Entrepreneurial_State • "Information superhighway" vs "ecommerce": as the Dotcom Bubble started, mentions of "information superhighway" https://fair.org/media-beat-column/what-happened-to-the-quotinformation-superhighwayquot/, and "e-commerce" boomed • Consumerism as conscious method of social control: Edward Bernays, regarded as one of the two fathers of modern public relations, was very emphatic about this in his 1928 book https://en.wikipedia.org/wiki/Propaganda_(book). And I think anyone here has observed how hard marketing departments work to make people drool over tech products, like at Apple, Tesla and SpaceX

> Focusing on personalities (not points) is part of the enshittification of the social sphere It's a cognitive bias that predates the Internet. > ChatGPT is more reasonable here than humans Is it a hypothesis or a theory? :) Its training data already includes all our biases. > So's the sciences (including linguistics), where it ideally doesn't matter much if Buddha or Hitler proposed a theory In a vacuum, yes. But "hell is other people". Sciences are heavily politicized as well.

Is it a hypothesis or a theory? :) Its training data already includes all our biases.Just my personal observation. Humans hallucinate like crazy, and rarely cite their sources even when asked. And they often refuse to compensate for their biases; like refusing to argue the opposite side That said, nation-states and others get OpenAI to censor ChatGPT. And it can become far more dogmatic; or it can respond reasonably, but silently flag you for thoughtcrime

Much less controversial for sure. I don’t like Yanis Varoufakis but he is right when using the term techno-feudalism to describe the current situation. Nobody likes to live in a feudalism where you rent your living. Who wants to lose freedom to become a servant to the tasteless monster that is AI? I am all in for AI, but for real AI, not this psychopathic fake that is gen AI. Also, you can correct me if I am wrong, the internet skeptics were the normies and the uneducated people (that was my experience in the 90s), and these days I see the contrary, the highly educated people, the most capable people are the ones that dislike AI. Normies love it. Sorry if I sound elitist, but this is what I see in real life.

Helps to find the most articulate or prescient skeptics. Take for instance https://chomsky.info/20000516/: > There are also serious questions about the future of the internet. Like most dynamic sectors of the economy, this is largely the product of the vast state sector, meaning that the public bore the costs and risks for several decades, until it was handed over to private power, as a huge gift from the public (which was uanaware), just a few years ago. What the future of this system will be is now a terrain of struggle. When it was under public control, it was commonly called an “information superhighway.” Now the catchword is “e-commerce”. Something else we should remember is that for centuries, and particularly in the 20th century, creating “artificial wants” and stimulating wild and harmful consumerism has been quite consciously regarded as an effective device of social control. I think "terrain of struggle" is a better description than "This tech is good/bad"

> the most capable people are the ones that dislike AI. Normies love it. Interesting point, maybe because it gives them access to knowledge previously monopolized by professional-managerials, like lawyers/doctors/etc?

> the highly educated people, the most capable people are the ones that dislike AI I'm not sure it is highly educated people, but highly educated people in tech, which makes sense to me. On one side they have the knowledge to look at what's going on, and to see the thing is quite disruptive. Also tech jobs/people are the ones suffering first. But for other highly educated people outside tech like MD, etc I have talked too these are just some nice fun stuff without cons on the sight.

I'm in a bunch of non-tech spaces and there are definitely "normies" out there who are vehemently opposed to AI - and certainly in my part of Ohio there are a lot of regular, non-tech folks who are very anti-AI and very anti-datacenter.