• 1 Post
  • 7 Comments
Joined 1 year ago
cake
Cake day: February 3rd, 2025

help-circle
  • @technocrit While I agree with the main point that “AI/LLMs has/have no agency”, I must be the boring, ackchyually person who points out and remembers some nerdy things.

    tl;dr: indeed, AIs and LLMs aren’t intelligent… we aren’t so intelligent as we think we are, either, because we hold no “exclusivity” of intelligence among biosphere (corvids, dolphins, etc) and because there’s no such thing as non-deterministic “intelligence”. We’re just biologically compelled to think that we can think and we’re the only ones to think, and this is just anthropocentric and naive from us (yeah, me included).

    If you have the patience to read a long and quite verbose text, it’s below. If you don’t, well, no problems, just stick to my tl;dr above.

    -----

    First and foremost, everything is ruled by physics. Deep down, everything is just energy and matter (the former of which, to quote the famous Einstein equation e = mc, is energy as well), and this inexorably includes living beings.

    Bodies, flesh, brains, nerves and other biological parts, they’re not so different from a computer case, CPUs/NPUs/TPUs, cables and other computer parts: to quote Sagan, it’s all “made of star stuff”, it’s all a bunch of quarks and other elementary particles clumped together and forming subatomic particles forming atoms forming molecules forming everything we know, including our very selves…

    Everything is compelled to follow the same laws of physics, everything is subjected to the same cosmic principles, everything is subjected to the same fundamental forces, everything is subjected to the same entropy, everything decays and ends (and this comment is just a reminder, a cosmic-wide Memento mori).

    It’s bleak, but this is the cosmic reality: cosmos is simply indifferent to all existence, and we’re essentially no different than our fancy “tools”, be it the wheel, the hammer, the steam engine, the Voyager twins or the modern dystopian electronic devices crafted to follow pieces of logical instructions, some of which were labelled by developers as “Markov Chains” and “Artificial Neural Networks”.

    Then, there’s also the human non-exclusivity among the biosphere: corvids (especially Corvus moneduloides, the New Caleidonian crow) are scientifically known for their intelligence, so are dolphins, chimpanzees and many other eukaryotas. Humans love to think we’re exclusive in that regard, but we’re not, we’re just fooling ourselves!

    IMHO, every time we try to argue “there’s no intelligence beyond humans”, it’s highly anthropocentric and quite biased/bigoted against the countless other species that currently exist on Earth (and possibly beyond this Pale Blue Dot as well). We humans often forgot how we are species ourselves (taxonomically classified as “Homo sapiens”). We tend to carry on our biological existences as if we were some kind of “deities” or “extraterrestrials” among a “primitive, wild life”.

    Furthermore, I can point out the myriad of philosophical points, such as the philosophical point raised by the mere mention of “senses” (“Because it’s bodiless. It has no senses, …”): “my senses deceive me” is the starting point for Cartesian (René Descartes) doubt. While Descarte’s conclusion, “Cogito ergo sum”, is highly anthropocentric, it’s often ignored or forgotten by those who hold anthropocentric views on intelligence, as people often ground the seemingly “exclusive” nature of human intelligence on the ability to “feel”.

    Many other philosophical musings deserve to be mentioned as well: lack of free will (stemming from the very fact that we were unable to choose our own births), the nature of “evil” (both the Hobbesian line regarding “human evilness” and the Epicurean paradox regarding “metaphysical evilness”), the social compliance (I must point out to documentaries from Derren Brown on this subject), the inevitability of Death, among other deep topics.

    All deep principles and ideas converging, IMHO, into the same bleak reality, one where we (supposedly “soul-bearing beings”) are no different from a “souless” machine, because we’re both part of an emergent phenomena (Ordo ab chao, the (apparent) order out of chaos) that has been taking place for Æons (billions of years and beyond, since the dawn of time itself).

    Yeah, I know how unpopular this worldview can be and how downvoted this comment will probably get. Still I don’t care: someone who gazed into the abyss must remember how the abyss always gazes us, even those of us who didn’t dare to gaze into the abyss yet.

    I’m someone compelled by my very neurodivergent nature to remember how we humans are just another fleeting arrangement of interconnected subsystems known as “biological organism”, one of which “managed” to throw stuff beyond the atmosphere (spacecrafts) while still unable to understand ourselves. We’re biologically programmed, just like the other living beings, to “fear Death”, even though our very cells are programmed to terminate on a regular basis (apoptosis) and we’re are subjected to the inexorable chronological falling towards “cosmic chaos” (entropy, as defined, “as time passes, the degree of disorder increases irreversibly”).


  • @gerryflap @bytesonbike

    many men are reluctant to make that step

    Sometimes it’s not the patient to blame. I made the step, countless times since my childhood… I sought help… Result? Got several, diverging diagnostics, several medications that didn’t work, until the most recent psychiatrist and psychologist some months ago: the psychiatrist said I got “nothing” (even when I had a fresh cut on my wrist) and the second “struggled to find any complaints from me”. So I simply gave up on seeking medical care (and “care” in general, human or whatnot). I don’t use AI for therapy because, as a former programmer, I’m deeply aware of their underlying Markov chain and NN algorithms, but sometimes their probabilistic outputs lead me to insights I couldn’t get from any living Homo sapiens beings (such as the possibility that I have “Geschwind Syndrome”, a condition of which will probably stay undiagnosed).


  • @Strawberry Governments and corporations are powerless to E2EE employed by the users themselves, such as GPG/GnuPG/PGP. What could/will UK gov do against GPG and similar tools, especially those which are open-source and freely available?

    I’m rooting for British people to defy their government and create their own pair of public and private keys using GPG/PGP or similar suite (preferably open-source, because they can be easily forked, adapted to easier UX/UI to any end-user, etc), sharing their public keys with each other so they can send enciphered messages, rendering useless such anti-E2EE British law.



  • @cyrano The “problem” (actually, the feature) with those censorship algorithms is that they rely a lot on the “exact contents” of the message (“Scunthorpe Problem”), so X is probably programmed to detect the Signal’s domain and block due to the presence of such link (similar to how Facebook was/is blocking links to the largest PixelFed instances, and then they also decided to block links to DistroWatch and official websites from various Linux distros), so it’s not programmed (yet) to censor just the “hexadecimal/base64/whatever” portion of the link alone. And there’s where Tox could shine: a handle is literally a hexadecimal sequence, without Tox’s domains, without URI Schemas, just a bunch of digits and letters from A to F.

    I don’t know why Tox isn’t mentioned as a “instant messaging platform for whistleblowers”: it got Onion (Tor) tunneling possibility (as well as tunneling it through I2P outproxies because it actually accepts any kind of SOCKS5 proxy), it’s registration-less (even Matrix needs registration) so it’s effectively anonymous IMO.

    SimpleX seems to be that, too, although I didn’t have the opportunity to use it more than I used Tox. But from the little I’ve used it, it’s similar to Signal in the sense that it’s a link (and a large link) and not simply a hash/hex sequence.


  • @scott I’m looking for something alternative to Github Gists, so I could quit it and go to a decentralized platform. Not the GitHub itself, but Github Gists, which is more like a multi-file “pastebin”.

    Something decentralized, federated, something akin to Writefreely where posts outside Writefreely (e.g., when seen from Mastodon) shows up as a shortened/summarized version, but tailored for code sharing, allowing for syntax highlighting, proper Markdown rendering and allowing for multiple text files within the same publication (“parts” within the “whole”, each one having their own filename; by “files”, I mean text files and source code such as .mjs, .js, .rb, .md, .txt, .csv, etc… not binary files such as images or movie-clips or audio or executables or zip/tarball/rar archives). Something that I could choose between hosting it myself (i.e., having my own instance) or using an existent instance, with the possibility of migration between instances (similar to Mastodon’s Export feature, but also allowing for importing exported publications).

    The closest thing would be @Daeraxa suggestion, Distbin (sorry Daeraxa for not getting to reply your comment yet, Friendica didn’t deliver it to me yet, but I saw your comment through Lemmy and I tested Distbin here). Sadly, as Daeraxa mentioned, that project is abandoned, with the last commit dating back from 6 years ago. Also, there are several issues (publications are anonymous because there’s no account system, publications can’t deal with multiple “files”, no deletion mechanism, etc). It’s the closest thing, but not the thing.

    Software forges (such as Vervis, Gitea and Forgejo) are really good for full projects. However, while I sometimes create full projects (where software forges, branches and forks, pull requests, CI/CD pipelines and source versioning would apply), the majority of my dev portfolio is composed of small-to-medium-sized code snippets experimenting with Node.js/Javascript, Python, Ruby and Shellscript, sometimes one-liner code snippets, and wrapping them up within full repos is overkill.

    Maybe I’m going to do a solution myself in the future, I’m not sure, it’s not exactly a plan or a promise. The core functionality seems simple to develop (looking at the Distbin’s source, it seems really simple) but integrating with ActivityPub and relaying info about multiple files within a publication in a format that’s compatible with other fediverse platforms (Mastodon, Friendica, etc) is a whole other story, in parts because I’m yet to learn about the technicalities of ActivityPub (i.e. how it really works behind the scenes beyond the /api/v1 calls).


  • I’m replying directly on the thread because one of the comments didn’t arrive in Friendica, yet (I’m seeing only 3 of 4 comments here on Friendica, I only could see the fourth comment through Lemmy).

    The suggested Forgejo (with public “instances” such as Codeberg) and Vervis seem good and promising. While they’re not exactly what I’m looking for (as Github Gists and Opengists are less than a software forge and more like a “Pastebin/Ghostbin”, they’re meant to allow sharing “repoless” code (they also have some Git integration, but there are no such things as branches, CI/CD, issue tracking, etc)), they’re useful to me, too, thanks everyone for suggesting them.