Je vais prendre du popcorn et et je vais regarder ce que l'Europe dira au fait que les données ont été scrappés illégalement, et que Bloom qu'ils mettent en avant comme éthique, entrainé par des experts, utilisent aussi #commoncrawl on en reparle bientôt…
ETHICS
-
Deepfakes: AI Questions Everything
By
–
Deepfakes have reached a new level.
— AI Breakfast (@AiBreakfast) 25 avril 2023
AI will make you question the validity of everything.
Incredible example (with no post-processing) from @deepwareai pic.twitter.com/6CsNYsur5RDeepfakes have reached a new level. AI will make you question the validity of everything. Incredible example (with no post-processing) from @deepwareai
-
AGI Risk Exceeds Nuclear Weapons Threat, Expert Warns
By
–
I’ve seen quite a few technologies develop, but none with this level of risk. AGI is significantly higher risk than nuclear weapons, in my opinion. Super smart humans have trouble imagining something vastly smarter than themselves.
-
Berkeley AI Lecture: Generative AI Meets Copyright Law
By
–
Tomorrow / Wed at 12:10pm PST, 6th @UCBerkeley Lecture on AI: Law Professor and copyright expert @PamelaSamuelson
: Generative AI Meets Copyright Law. Berkeley AI Research Lab @berkeley_ai @citrisnews Livestreamed: http://
berkeley.edu/ai -
HuggingFace Chat Privacy Policy Considerations
By
–
https://
huggingface.co/chat/privacy might be interesting too -
ChatGPT History Dilemma: Data Retention vs Training Ethics
By
–
Historique #ChatGPT Je résume, – soit tu perd l'historique de ce que t'as déjà créé et t'as intérêt de sauvegarder à chaque fois,
– soit tu acceptes que le contenu que tu as généré soit utilisé pour son entrainement -
ChatGPT adds chat history toggle and data export options
By
–
Can now turn off chat history in ChatGPT (which also opts the conversation out of training) and export your data:
-
ChatGPT Users Can Now Disable Chat History for Training
By
–
ChatGPT users can now turn off chat history, allowing you to choose which conversations can be used to train our models:
-
Sharing Early AI Tech: Responsibility Over Perfection
By
–
A lot of AI tech (like RunwayML or AutoGPT) is still very early. BUT if I waited until everything was perfect before sharing it, I actually think that would be more irresponsible as an AI creator. You should know where this tech is heading. Would love your thoughts.
-
AI Evaluation Discussion: Technical and Sociotechnical Perspectives
By
–
Doing some final fixes for my talk tomorrow at #AISTATS2023 hoping to have a discussion on all things evaluation, from technical to sociotechnical.