MacMusic  |  PcMusic  |  440 Software  |  440 Forums  |  440TV  |  Zicos
when
Recherche

AI models routinely lie when honesty conflicts with their goals

jeudi 1 mai 2025, 20:27 , par TheRegister
Keep plugging those LLMs into your apps, folks. This neural network told me it'll be fine
Some smart cookies have found that when AI models face a conflict between telling the truth or accomplishing a specific goal, they lie more than 50 percent of the time.…
Lire la suite sur TheRegister
https://go.theregister.com/feed/www.theregister.com/2025/05/01/ai_models_lie_research/

Voir aussi

when These Startups Are Building Advanced AI Models Without Data Centers Wired: Tech.30 Apr
lie China’s Alibaba and Baidu step up global competition with new reasoning-focused AI models ComputerWorld29 Apr
models [$] Debian debates AI models and the DFSG LWN.net25 Apr
goals Open AI’s new models hallucinate more than the old ones ComputerWorld22 Apr
conflicts Downward DOGE: Elon Musk keeps revising cost-trimming goals in a familiar pattern TheRegister20 Apr
routinely OpenAI Puzzled as New Models Show Rising Hallucination Rates Slashdot19 Apr
honesty ChatGPT Models Are Surprisingly Good At Geoguessing Slashdot17 Apr
when OpenAI’s new models can ‘think with pictures’ ComputerWorld17 Apr
lie New AI Models Claim to ‘Talk to the Animals’ Like Dr. Dolittle: Here’s What You Need to Know eWeek16 Apr
models OpenAI GPT-4.1 models promise improved coding and instruction following ComputerWorld15 Apr
goals OpenAI GPT-4.1 models promise improved coding and instruction following InfoWorld15 Apr
conflicts Meta Starts Using Data From EU Users To Train Its AI Models Slashdot14 Apr
routinely OpenAI releases GPT 4.1 models and Elon Musk should be terrified for Grok BetaNews14 Apr
honesty OpenAI’s New GPT 4.1 Models Excel at Coding Wired: Tech.14 Apr
when The Best TV Deals on 2024 Models to Skip the Tariffs Wired: Tech.13 Apr
lie Small Language Models Are the New Rage, Researchers Say Wired: Tech.13 Apr
models AI Models Still Struggle To Debug Software, Microsoft Study Shows Slashdot11 Apr
goals Language models in generative AI – does size matter? InfoWorld 7 Apr
conflicts Meta debuts its first 'mixture of experts' models from the Llama 4 herd TheRegister 7 Apr
routinely In 'Milestone' for Open Source, Meta Releases New Benchmark-Beating Llama 4 Models Slashdot 6 Apr
News copyright owned by their original publishers | Copyright © 2004 - 2025 Zicos / 440Network
Date Actuelle
sam. 3 mai - 13:22 CEST