@guerda@ruhr.socialI am looking for a website which existed a couple of years ago and listed use cases for #MachineLearning. Next to each use case, papers solving the task were listed including the f1 score or other success metrics.
It was extremely useful to see which approaches can be applied to each use case like document classification, image recognition,
@thomasfuchs@hachyderm.ioWhat really makes me irate about how LLMs are marketed and sold is that if these companies instead spend their time and money to make highly specialized versions for them we could have amazing and actually helpful tools without the ick.
This could both work much better for many use cases (for example for correlating documents and giving a list of results like a search engine instead of tedious palaver) and they wouldn't need to steal data (Professor Bender calls it succinctly "datasets too large to care")[1].
But they're pursuing "AGI" (which is provenly impossible to do with LLMs) and endless growth.
[1] https://dair-community.social/@emilymbender/116109627131276897
@vyskocilm@witter.cz> We found that the “Other” category increased to 11% this year, and this was primarily driven by Hetzner (20% of Other responses); we plan to include Hetzner as a response choice in next year’s survey.
https://go.dev/blog/survey2025
Well done @…
@hex@kolektiva.socialThe people building LLMs are trying to kill all humans. Literally. They are exterminationists who want to replace human life with AI and upload their brains to the cloud. They're AI death cultists.
They are the enemy. Anything that empowers them, at all, in any way, is unethical and suicidal.
I don't think admitting that there can be good use cases for LLMs does that, because the use cases are not being served by all this training. The *reason* they keep training is that they're trying to make LLMs do something that isn't possible.... because they're cultists.
That's the point, IMHO.
@adulau@infosec.exchangeHave you ever tried doing digital forensics using an SBOM or even just gathering evidence for a technical investigation from one?
No file hashes, a single cryptographic signature covering an arbitrary set of files, and often missing full paths or permissions.
Many SBOM standards need a serious revamp if they are to support DFIR use cases
#dfir
@nohillside@smnn.chApple might be in a better position than others to launch this successfully because it can build it on an established base (all iPhones out there). Still don‘t see the use cases though. And don‘t even mention privacy issues 🙄
Apple Developing AI Wearable Pin - https://apple.slashdot.org/s…
@cjust@infosec.exchange“No one knows right now what the right reference architectures or use cases are for their institution,” Dorian Smiley, CTO and founder of AI software engineering company Codestrap, told The Register.
“From the large language model perspective, people aren’t really addressing the fallibility of the underlying text,” CEO Connor Deeks added.
As software engineers continue to be put under pressure to use AI for…
@cellfourteen@social.petertoushkov.euI like AI. I like robots. I love machine learning automation. I don't like it when their use cases are replacing people, spying on or profiling people, prosecuting people, submitting people into subscription, creating "art", "videos", "pictures" and scumbag "memes", warfare, propaganda, disinformation, deepfakes of any kind, ads, trolling, pumping up stocks, just plain wrong search results that force you to waste twice as much time to confirm they …
@cdarwin@c.imThe British Home Office’s use of Artificial Intelligence (AI) in processing asylum claims could be unlawful,
legal experts have warned,
opening up the possibility of court action against the government.m
Caseworkers at the Home Office use AI to summarise interview transcripts with asylum seekers.
It is also used to search policy guidance, such as information about whether a country is deemed safe to return to.
Asylum seekers are not told when AI is used on thei…
@thomasfuchs@hachyderm.ioThe similarities between VR and AI are actually really striking; both are technologies that are amazing for a few narrow niche use cases and completely unsuitable for anything else.
But they are both heavily marketed and hyped exactly for the use cases where they suck and make no sense.
@shriramk@mastodon.socialI regularly use Claude Code, but I'm struggling to see much of use yet for Claude Cowork. (I've installed it just in case I think of a use, but other than giving me a quick keystroke to access Claude itself, not much has come to mind.) What are your interesting use-cases?
@Techmeme@techhub.socialFei-Fei Li's World Labs raised $200M from Autodesk, part of a larger round; the companies are exploring integrating World Labs' AI models in Autodesk products (Rebecca Bellan/TechCrunch)
https://techcrunch.com/2026/02/18/worl
@ErikJonker@mastodon.socialHow generative AI is used in the financial sector.
#AI
@cyrevolt@mastodon.socialSo yesterday I learned that #ping != ping.
https://github.com/u-root/u-root/issues/3361
Apparently, even just looking at the -w/-W option, which is different between
@tore@openbiblio.social#DataCite Schema 4.7 is available ⤵️
Supporting DOIs for Presentations and Posters
As suggested by community members, we’ve added new types for Presentations and Posters to the resourceTypeGeneral controlled vocabulary. These additions will make it possible to consistently distinguish these important output types that are integral parts of the research life cycle, helping harvester…
@bogo@hapyyr.comQ: How do you convince people to use privacy respecting technologies for communication?
I am looking for real life use-cases to include into my #FOSDEM talk and beyond.
If you have encouraged someone to switch to say Signal, or Sessions, or XMPP or some other tool/protocol, please share here or via DM.
Anonymity is guaranteed. You can use also my Sessions messenger session: 0531634…
@drbruced@aus.socialRE: https://cosocial.ca/@timbray/116087687347120369
The fact that this post admits the possibility of some use cases for LLMs in software development is sure to piss off lots of people, but there is plenty of coverage of the downsides to even it out. Well bala…
@migueldeicaza@mastodon.socialWhen you are a Swift godot user, every day is xmas!
https://forums.swift.org/t/pitch-emscripten-target-support-for-swift/85310
@ruth_mottram@fediscience.orgProud PhD supervisor moment: @… have nice write up to Elke Schlager's brilliant work, (now a preprint in @…) on how we can use #MachineLearning methods to emulate #Greenland ice sheet melt via European Weather Cloud computing
https://europeanweather.cloud/use-cases/machine-learning-emulation-accelerate-climate-science
@Techmeme@techhub.socialA survey of 1,692 US physicians: over 80% use AI professionally and the most common use cases are medical research summarization and clinical care documentation (American Medical Association)
https://www.ama-assn.org/press-center/ama-
@mgorny@social.treehouse.systemsWhenever people are commenting on another half-assed, crappy #LLM feat, claiming that there are "some" use cases for this "#AI", substitute "AI" with "genocide".
Because, you know, there are "use cases" for genocide too, and apparently a lot of people don't mind, as long as they can benefit from it and look the other way.
#NoAI
@chiraag@mastodon.onlineMakes me feel old that I remember having to tweak the `/etc/X11/xorg.conf` to get my display working. Nowadays, I mostly don't have to think about any of that anymore b/c Wayland "Just Works" (TM), at least for the cases that I use (simple laptop/desktop setup). I lost AwesomeWM and Xmonad of course, but Sway still scratches a similar itch. If I'm masochistic enough, I might try doing something on top of wlroots with config in Haskell 😂
@thomasfuchs@hachyderm.ioRE: https://hachyderm.io/@thomasfuchs/115883445350346364
I think the technology could be used responsibly for narrow use cases (e.g. writing code is a narrow use case), with the very important caveat that the training data needs to be properly licensed and authors properly compensated (unless they licensed stuff free for commercial use); as well as companies offering it being transparent and honest about what LLMs actually do.
Anyway, even if that’s the case (and that’s a big if and certainly not the case today) I have serious doubts concerning the long-term financial sustainability of companies like idk Anthropic.
@Marwe@troet.cafeDie EU-Kommision macht eine Konsultation zu #OpenSource für den Einsatz in ihrer Verwaltung und sucht Stellungnahmen und Use Cases, z.B. von Unternehmen, Verwaltungen oder Einzelpersonen:
> The European Open Digital Ecosystem Strategy will set out:
a strategic approach to the open source sector in the EU [...]
a strategic and operational framework to strengthen t…
@inthehands@hachyderm.ioTo some extent, that’s the job of abstractions:
Maybe I don’t want to have to decide exactly what a button should look like, so I use the OS’s provided Button widget.
Maybe I don’t want to have to decide exactly how integer addition should work, so I use the language’s integer type.
In both cases, part of what the abstraction provides is “just do this thing the standard way.” But…
@theodric@social.linux.pizzaGrapheneOS should include a threat model slider ranging from "idk I thought the OS name sounded cool" to "I am a target of nation-state intelligence agencies" which adapts the hardness to suit real-world use cases rather than treating everyone like they're 007 on assignment in 1983 Moscow
@ruth_mottram@fediscience.orgThe preprint of the paper is here btw: #MachineLearning methods to emulate #Greenland ice sheet melt via European Weather Cloud computing
https://europeanweather.cloud/use-cases/machine-learning-emulation-accelerate-climate-science
@cdarwin@c.imThe Trump administration’s immigration enforcement arm
is requesting unfettered access to what is considered to be the most comprehensive government database of people in the United States
and their most private information,
including sensitive details about individual children, according to six current and former federal officials.
It is called the "Federal Parent Locator Service", and it’s meant for finding people who owe child support.
Granting acc…
@wyri@toot-toot.wyrihaxim.us@… Interesting, there is no config at all on spegel? My mainn use cases would be pre-fetching certain images, and quicker start up times for runners when there is a new version. But it's a big nice to have tbh.
@karlauerbach@sfba.socialThe Feds keep saying "We did not" or "We will not" buy this or that empty warehouse for use as an ICE prison.
What they are hiding is that in order for a warehouse to be an ICE/CBP prison that the warehouse need not be "owned" by the Feds.
Rather the warehouse could be bought by a maga-friend, like Elon or Ellison etc and then leased to the government.
The Feds would not bey lying when they say they are not buying the warehouse. But they will …
@geant@mstdn.social🆔 Why join the GÉANT project’s Wallet subtask mailing list?
Digital identity wallets are becoming a reality across Europe, and the decisions made now will shape how students, researchers and institutions collaborate in the future.
Join the conversation, contribute R&E use cases, explore interoperability, and help ensure that the values of openness, trust, and collaboration remain at the heart of digital identity.
🔗 Subscribe here:
@Techmeme@techhub.socialHow the US government is adopting AI: NASA reports 420 AI use cases in 2025, up from 18 in 2024, HHS reports 398, Energy 325, DOJ 295, Interior 234, and DHS 205 (Washington Post)
https://www.washingtonpost.com/technology/2026/02/09/trump-administration-ai-p…
@cdarwin@c.imThe Department of Homeland Security has been quietly demanding tech companies
turn over user information about critics of the Trump administration, according to reports.
In several cases over recent months, Homeland Security has relied on the use of administrative subpoenas
to seek identifiable information about individuals who run anonymous Instagram accounts,
which share posts about ICE immigration raids in their local neighborhoods.
These subpoenas have also…
@cdarwin@c.imThe Department of Homeland Security
has been quietly demanding
tech companies turn over user information about critics of the Trump administration,
according to reports.
In several cases over recent months,
Homeland Security has relied on the use of administrative subpoenas
to seek identifiable information about individuals who run anonymous Instagram accounts,
which share posts about ICE immigration raids in their local neighborhoods.
These s…
@inthehands@hachyderm.ioMy solution to the problem @… mentions:
1. Make password the •only• login field. No username/email! Just password, and you’re in!
2. Because they now identify users, passwords must to be unique across all users in a system.
3. For security reasons, require passwords to be unique across all sites on the entire Internet. A password can be used at most once in human history.
4. This renders account recovery impossible in many cases.
5. Over time, users are thus able to use fewer and fewer sites, eventually being forced offline altogether.
6. Success!
https://mastodon.social/@jwz/115776511391611416
@Techmeme@techhub.socialSelf-driving software startup Oxa raised a $103M Series D, with $50M coming from the UK government's National Wealth Fund; Nvidia's NVentures also invested (Tom Nugent/Sifted)
https://sifted.eu/articles/oxa-raises-103m-nvidia-bp
@arXiv_csLG_bot@mastoxiv.pageSOM-VQ: Topology-Aware Tokenization for Interactive Generative Models
Alessandro Londei, Denise Lanzieri, Matteo Benati
https://arxiv.org/abs/2602.21133 https://arxiv.org/pdf/2602.21133 https://arxiv.org/html/2602.21133
arXiv:2602.21133v1 Announce Type: new
Abstract: Vector-quantized representations enable powerful discrete generative models but lack semantic structure in token space, limiting interpretable human control. We introduce SOM-VQ, a tokenization method that combines vector quantization with Self-Organizing Maps to learn discrete codebooks with explicit low-dimensional topology. Unlike standard VQ-VAE, SOM-VQ uses topology-aware updates that preserve neighborhood structure: nearby tokens on a learned grid correspond to semantically similar states, enabling direct geometric manipulation of the latent space. We demonstrate that SOM-VQ produces more learnable token sequences in the evaluated domains while providing an explicit navigable geometry in code space. Critically, the topological organization enables intuitive human-in-the-loop control: users can steer generation by manipulating distances in token space, achieving semantic alignment without frame-level constraints. We focus on human motion generation - a domain where kinematic structure, smooth temporal continuity, and interactive use cases (choreography, rehabilitation, HCI) make topology-aware control especially natural - demonstrating controlled divergence and convergence from reference sequences through simple grid-based sampling. SOM-VQ provides a general framework for interpretable discrete representations applicable to music, gesture, and other interactive generative domains.
toXiv_bot_toot
@Techmeme@techhub.socialA study finds GPT-5.2, Claude Sonnet 4, and Gemini 3 Flash deployed tactical nuclear weapons in 95% of 21 simulated war game scenarios, and never surrendered (Chris Stokel-Walker/New Scientist)
https://www.newscientist.com/article/25168
@Techmeme@techhub.socialChinese humanoid robot startup Galbot raised over $300M led by China Mobile's industry investment fund at a $3B valuation, bringing its total funding to $800M (Eudora Wang/DealStreetAsia)
https://www.dealstreetasia.com/stories/galbot-raises-over-300m-467245
