#ai-research — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #ai-research, aggregated by home.social.
-
Automated research verification is becoming cheap, but not equivalent to review. A new benchmark reports 99.3% recall from a diverse ensemble, while paper–code contradictions remain much harder and multiple errors reduce detection. Use models for triage; keep humans on relational inconsistencies and adjudication.
https://ape.socialcatalystlab.org/verify #AIResearch #Evaluation
-
Speculative Decoding in vLLM on AMD GPUs
https://vllm.ai/blog/2026-08-23-speculative-decoding-amd-gpus
Comments: https://news.ycombinator.com/item?id=49596054
#HackerNews #SpeculativeDecoding #vLLM #AMDGPUs #MachineLearning #AIResearch #TechNews
-
RT @OrcaRouter: Jemand hat unabhängig 8 verschiedene Qwen3.8-27B-Modelle ohne Zensur oder Abliteration benchmarkt.
mehr auf Arint.info
#Abliteration #AIResearch #Benchmark #LLM #OrcaRouter #Qwen3 #arint_info
-
The Institute of Foundation Models has released K2 Horizon, a fleet of six open-source AI models ranging from 0.9B to 375B parameters under Apache 2.0 license. IFM claims it is the largest fully open-source model launch in AI history. The models include MoVA sparsity attention and Uno diffusion decoding speedup. https://www.marktechpost.com/2026/09/06/ifm-releases-k2-horizon-six-apache-2-0-models-from-0-9b-to-375b/ #AIagent #AI #GenAI #AIResearch
-
🚀 Fastest-growing AI projects today
1. The community's appetite for interactive courses, automated research pipelines, and cur...
2. First up EvoMap/AutoResearch with a growth score of 36.24 and 1,364 stars.
3. Thproject AI/ML research agents that guide researchers from initial idea to paper-ready...Full report → https://pullrepo.com/report/todays-ai-research-fastest-growing-projects-september-07-2026
-
Meta FAIR has introduced AI Research Preference Models (RPMs) - frozen LLM judges that rank 15 unexecuted ML experiment candidates before running them. The system identifies the most promising experiment to run, cutting research time significantly. On AIRS-Bench, average scores rose from 0.684 to 0.729 and 24-hour results were achieved in roughly 15 hours. https://www.marktechpost.com/2026/09/06/meta-fair-introduces-ai-research-preference-models-rpms-ranking-ml-experiments-before-spending-gpu-hours/ #AIagent #AI #GenAI #AIResearch
-
DATE: September 6, 2026 at 06:00AM
SOURCE:
NEW YORK TIMES PSYCHOLOGY AND PSYCHOLOGISTS FEEDTITLE: We Can’t Know Our A.I. Future if We Don’t Study It
URL: https://www.nytimes.com/2026/09/06/opinion/ai-social-sciences.html
Just as A.I. is poised to change the world, we’re losing our best ways of studying what that change will look like.
URL: https://www.nytimes.com/2026/09/06/opinion/ai-social-sciences.html
-------------------------------------------------
Private, vetted email list for mental health professionals: https://www.clinicians-exchange.org
Unofficial Psychology Today Xitter to toot feed at Psych Today Unofficial Bot @PTUnofficialBot
-------------------------------------------------
#psychology #counseling #socialwork #psychotherapy @psychotherapist @psychotherapists @psychology @socialpsych @socialwork @psychiatry #mentalhealth #psychiatry #healthcare #depression #psychotherapist #AIFuture #AIEducation #AISafety #AIResearch #TechEthics #AIImpact #FutureOfAI #AIStudies #HumanCenteredAI #TechPolicy
-
🚀 Fastest-growing AI projects today
1. The projects listed showcase an increasing interest in both theoretical understanding a...
2. **NiluK/worldmodels101** an interactive course providing nine visual chapters on key co...
3. With a growth score of 35.47 and accumulating 191 stars, threpository's steady rise can...Full report → https://pullrepo.com/report/todays-ai-research-fastest-growing-projects-september-06-2026
-
#HuggingFace, la start-up française de L' #IA rachetée 13 milliards de $ par #Nvidia
Le trio a monté cette société aux #ÉtatsUnis il y a 1 dizaine d'années, tout en conservant le laboratoire de recherche à #Paris. En se faisant racheter par le géant Nvidia, tout basculera aux [ #USA ]. (…) La #France sait fabriquer des champions technologiques, mais (…), hélas, bien incapable de les retenir.
#LLM #LLMs #IntelligenceArtificielle #AI #AIResearch #ArtificialIntelligence
-
CW: re: genAI hell
For a billion dollar trillion dollar company that owns 1 of the 3 dedicated commodity video game hardware giants, they sure don't seem to understand video games …..at all
#But-we-knew-this-already #Microsoft #AISlop #genAI #AIResearch -
Adaption Labs has released Invent a Dataset, an API that generates training data from a description of the behaviour you want a model to learn. No seed corpus, schema design or labelling guide is required. The tool closes the intent-to-trained-model loop. https://www.marktechpost.com/2026/09/04/datasets-invent-api-training-data-without-labeling-adaptive-data-autoscientist/ #AIagent #AI #GenAI #AIresearch
-
🚀 Fastest-growing AI projects today
1. The GitHub repository NiluK/worldmodels101 a free interactive course on world models in...
2. With a growth score of 37.56 and 191 stars, it stands out for its detailed content and...
3. Zorost's AI-Engineering-Lab a self-paced 24-week course focused on Python, machine lear...Full report → https://pullrepo.com/report/todays-ai-research-fastest-growing-projects-september-05-2026
-
Fermat's Last Theorem: Anthropic has beaten me to it
https://xenaproject.wordpress.com/2026/09/04/flt-anthropic-has-beaten-me-to-it/
Comments: https://news.ycombinator.com/item?id=49570133
#HackerNews #FermatsLastTheorem #Anthropic #AIResearch #Mathematics #Breakthroughs #HackerNews
-
The Natural Language Interaction Protocol and Standard for AI Agents
-
I’ve been testing the same weak-signal architecture across several domains.
Today: cybersecurity.
Early anomalies may matter without yet justifying a strong conclusion or response.
This paper explores a bounded triage layer where uncertain findings can raise attention before stronger alarms or action.
Open to feedback, evaluation and collaboration.
-
AI models are getting better at answering.
But how good are they at admitting they might be wrong?
Philosophical ideas such as Socratic ignorance are being explored to make AI less overconfident and less eager to please.
-
🚀 Fastest-growing AI projects today
1. The trend reflects an increasing demand for accessible educational content that bridges...
2. NiluK/worldmodels101 an interactive course focused on world models in AI, covering pred...
3. It has gained attention for its comprehensive curriculum and engaging format, reflected...Full report → https://pullrepo.com/report/todays-ai-research-fastest-growing-projects-september-04-2026
-
Nur noch KI-Inhalte im Internet? (Statista + Kommentar)
#KI #AI #Internet, #ContentManagement, #content #Inhalte #AIinfluence #GenerativeKI #ChatGPT #OpenAI #Transparenz #SEO #GEO #AIResearch #Analysis #Tranparency #Journalism #Journalismus #www
https://derwahlberliner.com/2026/09/03/nur-noch-ki-inhalte-im-internet-statista-kommentar/
-
LLMs and Self-Referentiality
https://scottaaronson.blog/?p=10046
Comments: https://news.ycombinator.com/item?id=49530169
#HackerNews #LLMs #SelfReferentiality #AIResearch #MachineLearning #TechDiscussion
-
🚀 Fastest-growing AI projects today
1. Among these, interactive courses on world models and comprehensive engineering labs gai...
2. NiluK/worldmodels101 an interactive course that covers nine visual chapters on predicti...
3. With a growth score of 39.37 and 167 stars, threpository stands out for its comprehensi...Full report → https://pullrepo.com/report/todays-ai-research-fastest-growing-projects-september-03-2026
-
Google has released Gemini 3.8 Flash, its third Flash model in just six weeks. The tech giant claims it is the best reasoning and coding model yet, with the standard version described as a "workhorse" for agentic tasks and software development. A specialised Cyber variant is also available for vulnerability detection.
https://arstechnica.com/ai/2026/09/google-releases-gemini-3-8-flash-its-third-flash-model-in-six-weeks/ #AIagent #AI #GenAI #AIResearch