Skip to content

Piero Bosio Social Web Site Personale Logo Fediverso

Social Forum federato con il resto del mondo. Non contano le istanze, contano le persone

lol, "if only someone had warned us about this sort of thing?!"

Uncategorized
42 31 3

Gli ultimi otto messaggi ricevuti dalla Federazione
Post suggeriti
  • 0 Votes
    1 Posts
    8 Views
    Language models cannot reliably distinguish belief from knowledge and factAbstract-----------«As language models (LMs) increasingly infiltrate into high-stakes domains such as law, medicine, journalism and science, their ability to distinguish belief from knowledge, and fact from fiction, becomes imperative. Failure to make such distinctions can mislead diagnoses, distort judicial judgments and amplify misinformation. Here we evaluate 24 cutting-edge LMs using a new KaBLE benchmark of 13,000 questions across 13 epistemic tasks. Our findings reveal crucial limitations. In particular, all models tested systematically fail to acknowledge first-person false beliefs, with GPT-4o dropping from 98.2% to 64.4% accuracy and DeepSeek R1 plummeting from over 90% to 14.4%. Further, models process third-person false beliefs with substantially higher accuracy (95% for newer models; 79% for older ones) than first-person false beliefs (62.6% for newer; 52.5% for older), revealing a troubling attribution bias. We also find that, while recent models show competence in recursive knowledge tasks, they still rely on inconsistent reasoning strategies, suggesting superficial pattern matching rather than robust epistemic understanding. Most models lack a robust understanding of the factive nature of knowledge, that knowledge inherently requires truth. These limitations necessitate urgent improvements before deploying LMs in high-stakes domains where epistemic distinctions are crucial.»#ai #LLMs #epistemology #knowledgehttps://www.nature.com/articles/s42256-025-01113-8
  • 0 Votes
    1 Posts
    6 Views
    I visited @canodrom this morning. I forgot to leave some stickers… Yet another reason for me to come back to Barcelona as soon as possible. Lovely people, super curious projects and approach, especially in the relationship with el barrio (neoghborhood).It was interesting curious to learn that the whole Canodrom is manages by @colectic_coop, a cooperative that won the 4-years grant by the municipality of Barcelona. The vast majority of the funding is both public and local! #Canodròm #Canodrom #Canodromo #Barcelona #coop #cooperative #FeministTech #technology
  • 0 Votes
    10 Posts
    23 Views
    They really are the bad guy. It is absolutely wild what they are able to get away with.
  • 0 Votes
    1 Posts
    15 Views
    Have you heard of @nlnet? They financially support organisations & people that contribute to an open internet for all since 1997 (& they historically contributed to the early internet in Europe in the 1980s!).If you're working on a project that "helps fix the internet through open hardware, open software, open standards, open science and open data", you can apply for a grant on their website:https://nlnet.nl/#opensource #foss #oss #tech #technology #programming #coding #openinternet