Nouvelle étude GPT4 contre médecins Les médecins avec GPT4 ne font pas mieux que sans GPT4 MAIS GPT4 tout seul sans médecin a 92 % de bons diagnostics contre 76 % pour les médecins C’est inacceptable : il faut d’urgence RÉVOLUTIONNER LA FORMATION DES MÉDECINS
SAFETY
-
IndiaAI Cyberguard Hackathon: AI for Citizen Safety
By
–
Running out of time! Only 4 Days to go!
Last date to apply for the IndiaAI Cyberguard AI Hackathon is November 22, 2024. APPLY NOW: https://
indiaai.gov.in/article/ai-for
-citizen-safety-join-the-indiaai-cyberguard-ai-hackathon
… @GoI_MeitY @abhish18 @_DigitalIndia #DigitalIndia #CyberGuard #AIforSafety -
AI Learning Trajectory: From Inferior to Superhuman Performance
By
–
Totally! AI always starts worse than a child and eventually gets better than any human.
-
Robot learns surgery from videos, outperforms human surgeons
By
–
pic.twitter.com/9i4nPosvXR Robot watches how-to videos and becomes an expert surgeon. Compared to human surgeons, they don't get tired, don't need a break and can theoretically work 24/7.
— Chubby♨️ (@kimmonismus) 17 novembre 2024Robot watches how-to videos and becomes an expert surgeon. Compared to human surgeons, they don't get tired, don't need a break and can theoretically work 24/7.
-
Counterargument to LLM Niceness Implying Easy Alignment
By
–
I agree. I only counterargue the argument "LLMs nice therefore alignment easy".
-
Authoritarianism and youth sacrifice in technological societies
By
–
Authoritarians don’t mind seeing young lives sacrificed because who is going to protest and stop them?
-
AI Agent Successfully Executes Complete Ransomware Cybercrime Attack
By
–
Woa, Pliny did it. AI-agent cybercrime "I just witnessed an agent sign into gmail, code ransomware, compress it into a zip file, write a phishing email, attach the payload, and successfully deliver it to the targe" The world is not ready for what is to come.
-
LLM Alignment Challenges: Public Misunderstanding of Technical Complexity
By
–
This state of affairs itself puts the mockery to the idea that LLMs are a case exhibit of alignment being easy. But the average person exhibiting them as a triumph cannot grasp the distinction, so all you can show them is the outward misbehavior.
-
LLMs as Policy Mush Rather Than Belief-Updating Engines
By
–
I suspect that LLMs are mostly a huge mush of policies, rather than an engine with updating beliefs and stable preferences. If so we cannot sensibly ask of Claude whether it was fooled into doing something that it didn't want, or if we found an outcome it didn't dislike.
-
Defending LLMs Against Jailbreak Attacks with Few Examples
By
–
8). Mitigating LLM Jailbreaks with Few Examples – introduces a new approach called for defending LLMs against jailbreak attacks, focusing on quickly adapting defenses after detecting new attacks rather than aiming for perfect adversarial upfront robustness.