Good thing I never deleted my linkedin, that should be much cleaner than fediverse accounts

Given the stochastic nature of LLMs and the pseudo-darwinian nature of their training process, I sometimes wonder if geneticists wouldn’t be more suited to interpreting LLM output than programmers.

It is, but maybe they mean they want no limit whatsoever on post length.

which, well, if your instance starts sending out megabyte-sized text posts I don’t expect it to stay federated with many others for very long.

I see, thanks for the correction.

There used to be this website, but the url just loads up a scam site now (I’ve created this issue on the project’s tracker if anyone has additional info to contribute).

I don’t know how technical you are, @VieuxQueb@lemmy.ca , but you could try running the “defed-investigator” project locally.

lemmy.ml, no, but I’m fairly certain that lemmygrad.ml has been defederated from lemmy.world at least, if not others.

I’ll be honest, that “Iceberg Index” study doesn’t convince me just yet. It’s entirely built off of using LLMs to simulate human beings and the studies they cite to back up the effectiveness of such an approach are in paid journals that I can’t access. I also can’t figure out how exactly they mapped which jobs could be taken over by LLMs other than looking at 13k available “tools” (from MCPs to Zapier to OpenTools) and deciding which of the Bureau of Labor’s 923 listed skills they were capable of covering. Technically, they asked an LLM to look at the tool and decide the skills it covers, but they claim they manually reviewed this LLM’s output so I guess that counts.

Project Iceberg addresses this gap using Large Population Models to simulate the human–AI labor market, representing 151 million workers as autonomous agents executing over 32,000 skills across 3,000 counties and interacting with thousands of AI tools

from https://iceberg.mit.edu/report.pdf

Large Population Models is https://arxiv.org/abs/2507.09901 which mostly references https://github.com/AgentTorch/AgentTorch, which gives as an example of use the following:

user_prompt_template = "Your age is {age} {gender},{unemployment_rate} the number of COVID cases is {covid_cases}."
# Using Langchain to build LLM Agents
agent_profile = "You are a person living in NYC. Given some info about you and your surroundings, decide your willingness to work. Give answer as a single number between 0 and 1, only."

The whole thing perfectly straddles the line between bleeding-edge research and junk science for someone who hasn’t been near academia in 7 years like myself. Most of the procedure looks like they know what they’re doing, but if the entire thing is built on a faulty premise then there’s no guaranteeing any of their results.

In any case, none of the authors for the recent study are listed in that article on the previous study, so this isn’t necessarily a case of MIT as a whole changing it’s tune.

(The recent article also feels like a DOGE-style ploy to curry favor with the current administration and/or AI corporate circuit, but that is a purely vibes-based assessment I have of the tone and language, not a meaningful critique)

Chiming in to say: same, though it took a step further for me before I quit in disgust. I was ready to accept the api costs argument in good faith until I learned that a dev could not make a reddit client that would use my own api token. Which meant they didn’t (only) care about the api load, they care about ensuring that I see as many ads instead of posts that they can get away with.

Sadly, Google worsening their search results to juice their own (ad) numbers not long afterwards led to the general public learning about searching Reddit as a way to land on actual human-vetted info. Just as the core user base splintered and left in greater numbers than ever before, a tidal wave of new users joined and enthusiastically picked up the torch – without even realizing what they were contributing to.

I would love to read that study, as going off of your comment I could easily see it being a case of “more than 10% of jobs are bullshit jobs à la David Graeber so having an « AI » do them wouldn’t meaningfully change things” rather than “more than 10% of what can’t be done by previous automation now can be”.

I just searched on GitHub for "Sha1-Hulud: The Second Coming.": 692 repositories. On the first page of results I was able to find a repo clearly made by the malware, and in that repo I was able to find someone’s github token with a few applications of “decode from base64”.

This is pretty bad. I don’t know what exactly comes next, an awareness campaign to get people to clean their infected machines and packages?

The materials were also chosen with an eye on reducing maintenance and repetitive tasks for the yacht’s staff, so traditional materials like teak decks and wooden handrails are out, and composite alternatives are in. The diesel-electric power plant works alongside a battery storage system that allows Leviathan to operate for long stretches with no emissions, and it also features an advanced wastewater treatment system.

On the one hand, it’s nice that mr Newell seems to be reducing the footprint of their luxury yacht above and beyond most of what I have heard happens in the rest of the luxury yacht industry. On the other hand, I shudder to think of what the footprint for the manufacturing of this custom-designed, one-of-a-kind luxury yacht looked like. Not to mention ‘composite’ usually means some sort of plastic, so now there’ll be one more thing spewing microplastics directly into the ocean…

I want to chime in on the subject of community sidebars.

To my understanding, many of the mobile apps people use to interact with the fediverse (and more specifically the threadiverse) haven’t figured out a great way to render community sidebar content in a way that a new user knows that it exists. Sidebar content is accessible, but often hidden in a sub menu or a non-obvious interaction. I use Boost, for example; in it you swipe inwards from the right side of the screen to slide the sidebar into view. This isn’t surprising to me, a somewhat veteran Reddit user that expects communities to have sidebars and for those sidebars to be on the right side of the screen. However a user that doesn’t already know about community sidebars has almost no way of discovering their existence when they use Boost. Mobile apps have limited screen width so they tend to focus on their “principal use” (visiting a community to browse their posts), but if you don’t know that communities have sidebars in which they describe themselves and their posting and commenting rules it’s very easy to end up in OP’s position.

Not to excuse their comments nor question their ban; I agree with the decision by the mods of c/196 to not spend any more effort dealing with such an oblivious user.

I suspect many Lemmy clients are designed for experienced users who already know how to navigate the space(s) and how they function. Yet much of the “how do we introduce new people to the fediverse and onboard them?” discussions I’ve seen seem to settle on “suggest a generalist instance like LW or .zip, suggest a mobile app like Voyager, and make them start browsing! Newbs are put off by having to do work like read up on an instance”. I wonder how much this end up contributing to creating cases like OP’s.

Then again, !womensstuff@piefed.blahaj.zone was plagued for over a year by men claiming they were “just responding to posts in their /all feeds”. When told about the community’s rules and sidebar, the most common response was along the lines of “I can’t be bothered to read the community name before commenting on a post in my feed, now I need to navigate to the community and find their sidebar?? This community should find a way to prevent their posts from appearing in /all instead”. If these users aren’t going to the effort of reading the community name as displayed on posts then there’s no guarantee they would read community sidebars even if they were already on-screen, in front of their eyeballs.

Even in the comments on this post I can see the argument that basically boils down to “spaces that don’t cater to me should also bear the effort of keeping out of my way” being voiced.

In France we’ve had a few white trash rappers take a decidedly punk slant. Ptite Soeur and Gemroz came out with the album Kayfabe Chimera about a year ago ; the track “Kayfabe” is, to me, full of punk messaging. Femtogo has more recently collabed with Ptite Soeur and the album they released, Pretty Dollcorpse, also has a decidedly punk message.

In terms of “old-school” punk music, you might find better recommendations on mastodon /the blogiverse rather than here on lemmy/the threadiverse.

According to this article written in July, it’s a bit more dire than that if you take a step or two back. Basically, openai and their copycats/derivatives are being held up by investments from Microsoft, Google, Amazon, and Meta, who in turn are being held up by investments from Nvidia. If/when the whole chain collapses it’ll be more than 0.5% of earnings that disappear.

Quel manque d’imaginaire, n’empêche… T’es sur internet, t’as pas les contraintes habituelles de la télé, résultat : tu tournes une émission qui aurait sa place entre un épisode de Kohlanta et un de Toit le Monde Veut Prendre Sa Place.

J’en suis à me demander si c’est pas le seul canal restant pour un youtubeur qui a déjà tout fait pour assurer une certaine rentrée d’argent (sans tomber dans le traffic d’arnaques pour autant). Si c’est le cas, ça s’annonce très dur les fins de carrière de tous les autres youtubeurs…

Not that I disagree, just as someone who loves computers and programming it really feels like throwing out the baby along with the bathwater.

We could (should imo) be planning a sort of overthrow of the rich assholes who don’t share; make sure everyone has access to a computer, the electricity need to run it, and the knowledge to use it to their own benefit.

The second, longer quote in my previous comment is from the intro to a computer self-help/“how-to” book, Without Me You Are Nothing (pdf link).

“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.” – Frank Herbert

"Right now there is an explosive growth of the number of computers and things they can do. Not only are their numbers increasing at a dazzling rate, but the storage of information in giant data banks is growing in the same explosive way.

We have no way to control this now and none in sight. In fact, the very nature of this growth says that all controls will lag far behind computer developments. Any attempt to ban them will only drive com- puters underground. Never lose sight of the fact that computers “crunch time.” The speed at which computers can operate tells us that laws cannot keep up with them. The person with a computer can dance rings around you while you react as though you were embedded in molasses.

What can you do?

Get your own computer. Learn how to use it. We are here to help you make that first step: how to find the one that fits your needs and your pocketbook, where to put it, how to program it-all of the essentials. If you don’t do this, the Bill of Rights is dead and your individual liberties will go the way of the dodo." – also Frank Herbert

I hate how much we seem to be slowly careening towards Frank Herbert’s vision like the worse case of collective target fixation.

Incredible. All throughout my studies the “bitter lesson”, so to speak, was that analogue circuits just couldn’t hold a candle to digital ones in terms of reliability when operating on small currents, to the point that no one bothered to miniaturize in analogue anymore. Even guitar pedals are almost all transistor-based, because it’s so much more feasible to to manipulate small currents through binary, quantized signals than analogue ones (even though the analogue ones are theoretically infinitely more precise).

Here’s the publication in Nature: https://www.nature.com/articles/s41928-025-01477-0

Here is either the pre-print or an accompanying paper on arXiv: https://arxiv.org/pdf/2205.05853

I’m trying to figure out how they get around the compounding imprecision that is inherent to multiple analogue steps and actually manage to rival digital circuit’s precision; seems like a big part come from how they have managed to squash all of the useful “work” down into almost a single step thanks to clever use of operational amplifiers on the “edges” of their resistive random access memory array.

A side-effect of the li/unix ecosystem overall moving to wayland is that conky is no longer really an option for spicing up your desktop.

Without conky, it’s much less trivial to do “interesting” things beyond window tiling, window decorations, and/or color theming.

Thanks for sharing that link, I thought the watchy had been more-or-less abandoned by the community.

For what it’s worth, GitHub says that chronos firmware had it’s 1.0 release just 5 days ago, so I’m not surprised the other commenter didn’t know about it.