Wondering if I understood LLMs correctly. Like it is a stochastic parrot, you can use it to create texts that sound plausible.
But there isnt any sort of proofing algorithm.
Can any GPT actually look for grammar mistakes? spelling errors? what is important or the main message of a text?if any of the statements in a text are true?
if any of the sources exist?
Please advise with any papers that would show that using an LLM for any of those would work.
"For more than a century, scientific journals have been the pipes through which knowledge of the natural world flows into our culture. Now they’re being clogged with AI slop."
- Ross Andersen at The Atlantic
https://www.theatlantic.com/science/2026/01/ai-slop-science-publishing/685704/
#TheAtlanticMagazine #RossAndersen #AIslop #AIethics #AIresearch #SciComm #JournalPublishing #JournalEditing #PublicationEthics #ResearchIntegrity #ScienceEditing #ScientificMisconduct #ScienceFraud #AIinAcademia #AcademiaIA
"For more than a century, scientific journals have been the pipes through which knowledge of the natural world flows into our culture. Now they’re being clogged with AI slop."
- Ross Andersen at The Atlantic
https://www.theatlantic.com/science/2026/01/ai-slop-science-publishing/685704/
#TheAtlanticMagazine #RossAndersen #AIslop #AIethics #AIresearch #SciComm #JournalPublishing #JournalEditing #PublicationEthics #ResearchIntegrity #ScienceEditing #ScientificMisconduct #ScienceFraud #AIinAcademia #AcademiaIA
The briefing also features perspectives from:
👤 Prof. Dr. Hinrich Schütze, Ludwig-Maximilians-Universität München (LMU)
👤 Prof. Dr. Dorothea Kolossa, Technische Universität Berlin
👤 Dr. Paul Röttger, Oxford Internet Institute
👤 Dr. Jonas Geiping, Max Planck Institute for Intelligent Systems
📄 Read the full German briefing here:
https://sciencemediacenter.de/angebote/sprachmodelle-entwickeln-unerwuenschte-verhaltensweisen-26006
🧾 Nature paper:
https://www.nature.com/articles/s41586-025-09937-5
(4/4)
Most strikingly, she emphasises that just a few examples can cause far-reaching behavioural shifts in LLMs, potentially affecting current models as well. For practitioners, the takeaway is clear: careful training data curation and thorough testing after fine-tuning are essential.
(3/🧵 )
The briefing also features perspectives from:
👤 Prof. Dr. Hinrich Schütze, Ludwig-Maximilians-Universität München (LMU)
👤 Prof. Dr. Dorothea Kolossa, Technische Universität Berlin
👤 Dr. Paul Röttger, Oxford Internet Institute
👤 Dr. Jonas Geiping, Max Planck Institute for Intelligent Systems
📄 Read the full German briefing here:
https://sciencemediacenter.de/angebote/sprachmodelle-entwickeln-unerwuenschte-verhaltensweisen-26006
🧾 Nature paper:
https://www.nature.com/articles/s41586-025-09937-5
(4/4)
AI research is "Eating your own dog food" kind of field. It's both amusing and predictable.
Major AI conference flooded with peer reviews written fully by AI https://www.nature.com/articles/d41586-025-03506-6
"Controversy has erupted after 21% of manuscript reviews for an international AI conference were found to be generated by artificial intelligence. "
How did they find out, you ask? With the help of AI models, of course. "EditLens: Quantifying the Extent of AI Editing in Textmodel." https://arxiv.org/abs/2510.03154
AI research is "Eating your own dog food" kind of field. It's both amusing and predictable.
Major AI conference flooded with peer reviews written fully by AI https://www.nature.com/articles/d41586-025-03506-6
"Controversy has erupted after 21% of manuscript reviews for an international AI conference were found to be generated by artificial intelligence. "
How did they find out, you ask? With the help of AI models, of course. "EditLens: Quantifying the Extent of AI Editing in Textmodel." https://arxiv.org/abs/2510.03154
This website illustrates nicely how the US lost the competition–at least for now–in open(ish) LLM models: https://www.atomproject.ai/.
What it fails to mention is that the new open(ish) models are, more or less, on par in their capabilities with closed models as well.
It'll be interesting to see what the consequences are, but my money is on the open models fostering innovation, and not on closed models winning the race for AGI. #AIResearch #AGI_hype
/via Wired
AHAAHAHAHAHAAAA !! LOL !
Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models
This website illustrates nicely how the US lost the competition–at least for now–in open(ish) LLM models: https://www.atomproject.ai/.
What it fails to mention is that the new open(ish) models are, more or less, on par in their capabilities with closed models as well.
It'll be interesting to see what the consequences are, but my money is on the open models fostering innovation, and not on closed models winning the race for AGI. #AIResearch #AGI_hype
/via Wired
Well, that's a relief, isn't it? The Zuckerberg private foundation, sorry, the we-do-something-philanthropic-thingy-see-we-are-good-people-do-you?, stopped investing in diversity (bah!) and is now solving all our problems. How? With AI!
Whereas our colleagues publish study after study where in research applying AI does or does not make sense, some people still invest blindly in it. Or not?
Make no mistake: This is a well-founded investment! If you pour money with your business into little ROI, then putting money into something that creates apparent demand is a marketing investment.
Well, that's a relief, isn't it? The Zuckerberg private foundation, sorry, the we-do-something-philanthropic-thingy-see-we-are-good-people-do-you?, stopped investing in diversity (bah!) and is now solving all our problems. How? With AI!
Whereas our colleagues publish study after study where in research applying AI does or does not make sense, some people still invest blindly in it. Or not?
Make no mistake: This is a well-founded investment! If you pour money with your business into little ROI, then putting money into something that creates apparent demand is a marketing investment.
🔗 Learn more:
• Official website → https://sbert.net/
• Original paper → https://aclanthology.org/D19-1410.pdf
• GitHub repository → https://github.com/UKPLab/sentence-transformers
📰 Read the full announcements:
TU Darmstadt Press Release
→ https://www.tu-darmstadt.de/universitaet/aktuelles_meldungen/einzelansicht_528832.de.jsp
Hugging Face Blog Post
→ https://huggingface.co/blog/sentence-transformers-joins-hf
(2/2)
#UKPLab #HuggingFace #SentenceTransformers #NLP #AIresearch #OpenSource 🚀
AI Bros: "It's getting better all the time, soon it won't make mistakes, really!"
OpenAI, makers of ChatGPT: We just proved mathematically that "hallucinations" are inevitable.
Kunstlig intelligens
