I can't just

[ginevracat@toot.community]

@researchfairy I feel this too. Like the whole world is sucked into a collective Emperor's New Clothes psychosis and I'm yelling "You're all naked. Can't you feel the cold!" But now the shops only sell the new non-clothes too!

[oliverboehme@troet.cafe]

@researchfairy Yes we have to actively fight it

[tuomov@mastodon.social]

@researchfairy I’m tired of reading manuscripts full of general strategies and frameworks and repeating the same discussion ten times. If it smells like LLM, it is LLM.

That said, I’m tired of reading English without articles as well, and with sentences as long as one uses in Finnish. So, pick your poison. ️

[mikill@mastodon.world]

@researchfairy "is a major driver of climate change"? Currently or (potentially) in the future?

[kleines_z@queer.group]

@tanghus @researchfairy
Even if I use the DuckDuckGo NoAI search, I notice when I search for information, many results are websites with LLM-generated text. I notice after I read some stuff on several websites and then recognize the bits of text that are stolen from those websites and put together to a text that obviously wasn't understood by it's author, because it combines texts parts that contradict each other.

[robotistry@fediscience.org]

@GinevraCat @researchfairy Today I learned that some people in robotics are following the horrendous path laid out in a conference keynote a few years ago of "The robots seem to be more robust when we replace <hand-coded / assembled set of tools> with <big blob of machine learning>. Maybe we should be considering developing <end-to-end blob of machine learning> to replace all the perception and actuation and planning and control components!"

I attended this talk from a seat in the back with a conference-provided sandwich. I heard this and was silently screaming "Noooooo!" as I thought about how hard it is to verify and test an autonomous system when you *have* transparency and insight into the individual components and how they relate to each other.

If all you have is a blob of weights and a pointer to a model card and data sheet for the training data? How do you figure out what to test? How do you figure out what your test results mean in terms of evidence that the system will or won't work under operational conditions? It's a catastrophically bad idea in so many ways. And yet people are chasing lemmings off cliffs after it.

[kleines_z@queer.group]

@tanghus @researchfairy
Well, it says it does try to blacklist slop sites. But it's obviously hard to do.

[researchfairy@scholar.social]

@mikill Why are you nitpicking this?

[a_cubed@mastodon.social]

@researchfairy @tante
I resemble this remark. At least one student per semester submitting something without even hiding that it's LLM Bullshit. Not sure howmajy are using it and hiding it well instead oflearning to walk sk they can later run (I'm teaching writing skills at ug and pg. levels)

[mathew@universeodon.com]

@tanghus @researchfairy uBlacklist. https://ublacklist.github.io/

Works with DuckDuckGo and others as well as Google.

[quasit@kolektiva.social]

@tanghus @researchfairy

We need the option to create, share, and curate blacklists!

[shadowfetchai@mastodon.social]

@researchfairy Yeah, I feel that. It's like the App Store review process; I can follow all the rules, but I'm still subject to the downstream effects of other developers spamming, scamming, or just shipping garbage. It lowers the bar for everyone.

[hellomiakoda@pdx.social]

@researchfairy It's a very sad situation. AI has some very helpful uses, from finding errors in pagues upon pages of config very quickly, to helping those who are disable, etc... But the WAY it's being used, and the way it's being shoved in to absolutely everything, and the scale of it all, the AI situation we have right now is so incredibly destructive.
It's like harnessing radiation. We can make energy cleaner than coal and treat cancer, or we can nuke people. ...we nuked people

[jaypatelani@bsd.network]

@researchfairy you can always choose NetBSD. NetBSD foundation rejects AI code commits. More funds are needed to such foundations. While Linux graciously allows AI

[bpl@snac.bsd.cafe]

I would not look at this as 0-1 situation. If someone opposes a thing, one can not use it and loudly say about the reasons. It does not mean that everything around will change, but we need living examples that life without harmful things is possible. This can be inspiring to others. If people would have to stop using AI right now, the ban would have to come from autocrat - which is fine, but people would not understand why this change is needed.

[npars01@mstdn.social]

@researchfairy @tante

The billionaires promoting AI are some of the worst people on the planet.

Wannabe despots just like Putin, Koch Network, & #PrinceBonesaw. The greed.

All assuring us of their benign intent while frying the planet, eroding democracy & law, evading taxation, & creating huge mass layoff schemes

https://www.theguardian.com/commentisfree/2026/may/19/billionaires-ai-complacency-resistance

https://www.bloomberg.com/features/2026-new-ai-billionaires-list/

https://www.cnbc.com/2026/05/20/jeff-bezos-taxes-ai-corporations-trump.html

https://www.benzinga.com/markets/tech/26/05/52681853/bernie-sanders-ai-moving-too-fast-billionaires-congress

https://www.gadgetreview.com/silicon-valleys-soothing-lies-about-your-ai-future

https://www.theatlantic.com/technology/2025/12/ai-boom-young-billionaires/685360/

https://www.nytimes.com/2025/12/29/technology/new-billionaires-ai-boom.html

[evan@cosocial.ca]

@researchfairy AI is not a major driver of climate change.

All IT is responsible for about 2% of greenhouse gas emissions. AI is responsible for about 0.1% of greenhouse gas emissions.

You can know the actual major drivers of climate change. They are published annually by most countries, states, cities and companies. They are things like gas cars, cement manufacturing, rice cultivation, cattle.

I'm not arguing for AI; I'm arguing against not understanding climate change.