"The marginal risks of open models have been shown to not be as extreme as many people thought (at least for text only — multimodal is far riskier)." What makes multimodal riskier – assuming you mean vision is it concerns over surveillance, facial recognition etc?
ETHICS
-
Robot Form Study: How Appearance Affects Human Interaction Behavior
By
–
We tested a robot form, human-like character, animals, and objects. People change how they interact depending on looks. RTT Test arxiv.org/abs/2507.10812 Full set of human behaviors ieeexplore.ieee.org/document… We are inspired by @mpshanahan's essays and others. Feedback welcome!
→ View original post on X — @petitegeek, 2025-08-06 13:14 UTC
-
Robot Responses to Physical Stimuli and Gestural Communication
By
–
…such as kicking (pain), and other affective stimuli 4. The robot's responses to higher level emblems, eg. waving or heart gestures 5. The robot's responses to pointing to things in context, like clothing 6. The robot's responses to roleplay, especially based on its looks
→ View original post on X — @petitegeek, 2025-08-06 13:14 UTC
-
Testing Robot’s Sensing, Motor Abilities and Emotional Responses
By
–
In our preliminary experiments, people tested: 1. The robot's sensing abilities, such as whether it could track them visually or respond to poking 2. The robot's motor abilities, or whether they could copy their movements 3. The robot's responses to smiling, aggression,…
→ View original post on X — @petitegeek, 2025-08-06 13:14 UTC
-
React to This Test: Identifying Teleopererated Robots Through Movement
By
–
You have one minute to interact with a robot through movement only. At the end, you need to say whether it's tele-operated or not. That's the React to This (RTT) Test – one of the robot Turing tests presented at activeself.de/movement-matte… Here's how initial tests went down 👇
→ View original post on X — @petitegeek, 2025-08-06 13:14 UTC
-
Using AI Well: 3 Essential Steps to Consider
By
–
You’re Using AI — But Are You Using It Well? 3 Steps To Consider
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @demishassabis @Ronald_vanLoon @TamaraMcCleary @geoffreyhinton @goodfellow_ian @jeffdean @erikbryn http://
ow.ly/253630sOj7h -
LLMs Cannot Accurately Describe Their Own Capabilities
By
–
My guess is that the fraction of humans who understand enough about LLMs to ask those questions overlaps pretty heavily with the fraction of humans who know that LLMs are almost universally terrible at accurately answering questions about themselves
-
Lesson 5: Are You Ready for the AI Act?
By
–
𝐋𝐞ç𝐨𝐧 𝟓 𝐄𝐩𝐢𝐬𝐨𝐝𝐞 𝟏 : Ê𝐭𝐞𝐬-𝐯𝐨𝐮𝐬 𝐩𝐫ê𝐭 𝐩𝐨𝐮𝐫 𝐥’𝐈𝐀 𝐀𝐜𝐭 ?
Avant de plonger dans la rivière IA, pose ta bouée et lis bien le nouveau règlement. L’IA Act, c’est comme repérer les dangers dans ta rivière estivale : certains passages sont interdits, -
AI Outsourcing Threatens Deep Reasoning Skills Generation
By
–
Remembering facts is largely redundant and a large part of it can be done away with. However outsourcing reasoning to ChatGPT is the real danger. I worry too that the new generation is not indulging in deep reasoning and problem solving.
-
Swedish PM Admits Using AI for Second Opinion on Decisions
By
–
Le Premier ministre suédois a admis utiliser des d’IA pour obtenir un « second avis » sur ses décisions. Et ce n'est que la partie visible…