This is so tame tho

It does say “BIOS Setup Utility” at the top tho…

Meh, Israel does often get neutrally framed headlines, but “beat to death” is imho worse sounding than “killed” as it inherently implies a kind of barbaric character that shooting someone does not. Its not like those “died from …” where the perpetrator is left unnamed. Everything is there in this case: WHO / HOW / WHAT = ISRAELI SETTLERS / BEAT / TO DEATH

ARM devices kinda dont have a bios, so sadly this wouldnt be possible :(

Its nuts to me that we figured out their strategy like 10 years ago and people just chose to ignore that information and kept using all these platforms anyway.

Probably the biggest victim of the LLM hype is academia. I dont have anything quantitative to back this up, but the students around me that rely on chatgpt a lot seem to be failing their exams more than those that arent.

In case you do end up caring, have some technological horrors for your entertainment: https://youtu.be/NuIMZBseAOM?t=1972

There probably arent even any “participants” behind those bots. The AI companies have exhausted all the written content on the internet, so now they are scraping video and audio data. I wouldnt be surprised if that is exactly what those bots are doing. Either thats their primary job, or their secondary job that the actual participants dont even know about.

Cloud computing is slowly crumbling too. At least in germany more and more companies are going back to on prem, because there are just a bunch of studies showing that cloud stuff is more expensive, less flexible and often actually has more downtime. But unlike other stupid projects you cant just reverse the move to cloud in a few months, it will take years to unfuck everything.

This one is a very basic CSAM scanner that goes through lemmy image storage and just deletes stuff it deems bad. https://github.com/db0/fedi-safety I havent tried it tho, so i cant attest to its quality.

Im sure there are tools made for mastodon too, since it has a lot more users.

Oh yeah for sure there are plenty of great uses for machine learning, i would never deny that.

We arent really in disagreement tho. I am absolutely empathetic with their situation and their burnout. Im just saying they should have seen it coming. Those two things arent contradictory. I just find it odd how one can invest so much time into a public project, without being aware of the realities of having to moderate it. Im thankful for the people that host instances, but i will never do it, because i know how much of a pain it is.

This really is a complicated problem, but if the fediverse wants to grow without relying on slave labor for moderation like Meta and the rest, then we have to find ways to lighten the load on moderators. Thats why creating transparent pre moderation tools like the image scanners used by many fediverse instances is so important.

Idk why you are being downvoted. It is a real possiblity, but i just dont think its likely.

reliable and cheap

Because it simply wont ever be those things. LLMs become more expensive the better you try to make them, exponentially so. We will likely never see LLMs that are significantly better than what we have today. Basically all written human texts on the internet have been used for LLM training. Any further gains can only come from optimizing training algorithms and there hasnt been any real progress there for years now.

When you moderate a platform that the public is free to upload anything to, you will see everything. From insults to threats to gore to CSAM. Thats how the internet is. That is the worst part in my opinion, because you cant just look away, but actually have to deal with it.

LLMs have a more mystical pull than other previous hypes, but look at what happened with crypto and blockchain. Every fucking government and company was saying that its the future and now nobody gives a shit about those anymore. It might be later than you and i would like, but eventually it will pass.

Sounds like someone wasn’t prepared for the reality of human interaction on the internet.
I mean its completely fair to say that they don’t want to deal with it, but also what did they expect?

Nah, thats just a side effect. The primary purpose of AI is being a hype object that companies can use to inflate their stock. I like how this blogpost explained it https://pluralistic.net/2025/06/30/accounting-gaffs/

The labor market effect will be quite short lived. LLMs as a replacement to human labor will be gone before long simply because they will ruin any company, government or project that relies on them. Its a sort of natural selection.

ICEBlock is very suspicious, no claims they make about privacy can be verified and Apple has access to all the userdata. As it is centrally run, there is also a good chance of the dev being threatened into handing over data to the feds.

This is very relatable to me. When you are in too deep, you cant just drop it.