𝕀𝔸 : 𝕝𝕖 𝕘𝕣𝕒𝕟𝕕 𝕕é𝕣𝕒𝕚𝕝𝕝𝕖𝕞𝕖𝕟𝕥 La grande question n’est plus seulement "comment aligner les intelligences artificielles"… Mais sommes-nous nous-mêmes encore alignés avec ce que nous construisons ? Bienvenue dans Mes Flashs de la Semaine #37
SAFETY
-
OpenAI Board Rejects Altman’s AGI Nonprofit Ownership Claim
By
–
Sam Altman dit : « L'AGI, une fois développée, appartient à l'organisation à but non lucratif pour le bien de l'humanité » Les Membres du conseil d'administration d'OpenAI lui répondent : "NON" C’est complètement fou.
-
Image Generation Models Lose Detail Through Iterative Processing
By
–
it has a tendency to gradually move towards highly desaturated color palettes. most prompts turn to mush after enough generations
-
AryaXAI Launches AI Alignment Labs in Paris and Mumbai
By
–
Introducing the AryaXAI AI Alignment Lab in Paris & Mumbai !
We're doubling down on AI interpretability & alignment with new R&D hubs focused on:
– Scalable explainability
– Risk-aware alignment
– Responsible AI innovation
Stay tuned as we shape the future of #ResponsibleAI -
Columns cause harm debunking AI risks requires focused effort
By
–
You're right I misunderstood there. Unfortunately the harm you cause with these columns is too high & the amount of time we have to debunk everything point by point constantly is too limited compared to the deluge and the actual things we need to do to combat harms of AI.
-
Preventing Harms from Misbelief in AI Consciousness
By
–
Consciousness is a fascinating topic. But personally, I'd rather resources be directed towards preventing (human) harms coming from people mistakenly believing an AI system is conscious.
-
The Urgency of Interpretability: Understanding AI Models
By
–
The Urgency of Interpretability: Why it's crucial that we understand how AI models work
-
Model Welfare: Scientific Uncertainty and Research Approaches
By
–
We remain deeply uncertain about the idea of “model welfare”. There’s no scientific consensus on it—or even on how to research it. We’re approaching the topic as carefully as we can. Find out more:
-
AI Models and Subjective Experience: New Research Program
By
–
As AI models become more complex and more capable, is it possible that they’ll have experiences of their own? It’s an open question. We recently started a research program to investigate it.
-
Prompt Evaluation and Optimization: Continuous Iteration
By
–
5/ Evaluation & Optimization Prompts aren’t “set it and forget it.” They A/B test, evaluate, and refine. They track hallucinations, latency, cost, and safety. Prompt engineers treat prompts like software always in iteration.