Tootfinder

We get so easily fooled because the typical answer is often in fact the correct one. That’s a heuristic we use in our interactions with other humans — “That’s what people say!” — and an LLM using human social forms triggers that psychological process. That’s how this magic trick works.
2/

'The No. 1 thing on my mind': Harbaugh, Ravens seeking answers during the bye https://www.espn.com/nfl/story/_/id/46596240/nfl-baltimore-ravens-harbaugh-bad-start-bye-week

ReTraceQA: Evaluating Reasoning Traces of Small Language Models in Commonsense Question Answering
Francesco Maria Molfese, Luca Moroni, Ciro Porcaro, Simone Conia, Roberto Navigli
https://arxiv.org/abs/2510.09351

ReTraceQA: Evaluating Reasoning Traces of Small Language Models in Commonsense Question Answering
While Small Language Models (SLMs) have demonstrated promising performance on an increasingly wide array of commonsense reasoning benchmarks, current evaluation practices rely almost exclusively on the accuracy of their final answers, neglecting the validity of the reasoning processes that lead to those answers. To address this issue, we introduce ReTraceQA, a novel benchmark that introduces process-level evaluation for commonsense reasoning tasks. Our expert-annotated dataset reveals that in a…

Man könnte meinen, da mischt sich einer in die Souveränität der EU ein. Gerade Russland mag so etwas doch eigentlich gar nicht. #GespalteneZunge #Medewedew #Honk

Ukraine-Krieg im Liveticker: +++ 11:06 Medwedew droht EU mit Verfolgung wegen eingefrorener Vermögen +++
Lesen Sie alle aktuellen Entwicklungen und Nachrichten über den russischen Angriffskrieg auf die Ukraine im Liveticker von ntv.de.

VizCopilot: Fostering Appropriate Reliance on Enterprise Chatbots with Context Visualization
Sam Yu-Te Lee, Jingya Chen, Albert Calzaretto, Richard Lee, Alice Ferng, Mihaela Vorvoreanu
https://arxiv.org/abs/2510.11954

VizCopilot: Fostering Appropriate Reliance on Enterprise Chatbots with Context Visualization
Enterprise chatbots show promise in supporting knowledge workers in information synthesis tasks by retrieving context from large, heterogeneous databases before generating answers. However, when the retrieved context misaligns with user intentions, the chatbot often produces "irrelevantly right" responses that provide little value. In this work, we introduce VizCopilot, a prototype that incorporates visualization techniques to actively involve end-users in context alignment. By combining topic …

Han forteller at folk gikk ut i gatene på lŸrdag, uten den vedvarende frykten de har hatt i to år av plutselig å bli truffet av en kule, eller havne midt oppi et angrep eller eksplosjon.
– Vi er selvsagt lykkelige og forventningsfulle, og vi gratulerte hverandre. Men vi er også engstelige, så det ble ingen stor feiring, sier Ferrero.
De tar nemlig ingenting for gitt, de har sett avtaler bli brutt mange ganger fŸr.

Prest i Gaza: «Første natt uten bomber var veldig spesielt»
Carlos Ferrero forteller om livet i Den hellige families kirke etter våpenhvilen. 450 flyktninger, hvorav 170 barn, har bodd i kirken under krigen. Nå kan de endelig gå ut uten frykt.

The U.S. Navy failed to alert the public to high levels of airborne radioactive material detected almost a year ago at the Hunters Point Naval Shipyard in San Francisco
According to a notice warning community groups in neighborhoods around the defunct base,
the Navy notified the San Francisco Department of Public Health only this month about elevated levels of plutonium-239 found last November.
Community groups and at least one San Francisco supervisor called the 11-month…

Navy Took 11 Months to Alert SF to Airborne Plutonium at Hunters Point Shipyard Site | KQED
San Francisco officials and advocates are demanding answers after the U.S. Navy failed to alert the public to high levels of airborne radioactive material found last November.

Die #Klimakrise verweilt in den französischen Bergen:
Wegen fehlendem #Schnee und unsicherer Winter durch den #Klimawandel baut das französische Skigebiet Céüze nun dauerhaft seine

Kein Schnee mehr wegen Klimawandel: Skigebiet in Frankreich macht dicht
Bei idealen Bedingungen sind hier große und kleine Skifahrer auf der Hunau in Bödefeld unterwegs.Fotos vom Januar 2025Hunaulift Flutlicht Skischule Lift Ankerlift

Shallow Robustness, Deep Vulnerabilities: Multi-Turn Evaluation of Medical LLMs
Blazej Manczak, Eric Lin, Francisco Eiras, James O' Neill, Vaikkunth Mugunthan
https://arxiv.org/abs/2510.12255

Shallow Robustness, Deep Vulnerabilities: Multi-Turn Evaluation of Medical LLMs
Large language models (LLMs) are rapidly transitioning into medical clinical use, yet their reliability under realistic, multi-turn interactions remains poorly understood. Existing evaluation frameworks typically assess single-turn question answering under idealized conditions, overlooking the complexities of medical consultations where conflicting input, misleading context, and authority influence are common. We introduce MedQA-Followup, a framework for systematically evaluating multi-turn rob…

Anker paid users of its Eufy security cameras $2 per video of staged or real package and car thefts to train its AI systems from December 2024 to February 2025 (Lorenzo Franceschi-Bicchierai/TechCrunch)
https://techcrunch.com/2025/10/04/anke

Anker offered Eufy camera owners $2 per video for AI training | TechCrunch
Hundreds of Eufy customers have donated hundreds of thousands of videos to train the company’s AI systems.