Complètement d’accord Hervé mais c’est aussi une réalité qu’il faut avoir en tête. Les Américains sont très en avance sur ces sujets. D’où l’importance de mettre aussi des gardes fous !
SAFETY
-
OpenAI partners Los Alamos, AI recreates vision, BNP Paribas backs Mistral
By
–
⦂ #OpenAI s’associe à Los Alamos pour tester les menaces biologiques Une IA recrée ce que les gens voient Pourquoi @BNPParibas s’allie à @MistralAI ? @_sophieheller nous explique tout en
-
Claude’s Limitations in Structured Storytelling and Narrative Frameworks
By
–
I like this idea, but creativity is only part of the problem here. Claude is also worse at following some kind of bedtime story framework like the hero's journey. I would expect it to give me at least a bland, but more structured story.
-
Claude’s Limitations with Narrative Story Structure
By
–
Yes, it's true with self-merges too. But I'd say it doesn't explain why Claude is also worse at following the hero's journey or some kind of traditional bedtime story framework.
-
Alignment Tax Impact on AI Model Performance Surprises
By
–
It's possible but still kind of surprising considering that bedtime stories are already "ultra-safe." We know about the alignment tax, but I didn't expect it to make the model that much worse.
-
Producer Liability for GenAI Systems Development and Compliance
By
–
Here's what adding teeth to (2) looks like @IATSE
: (ii) Producer is responsible & liable for ensuring GenAI systems are developed in non-infringing ways, respecting privacy, and without illegal content (e.g. CSAM). This obligation may not be offloaded through contractual means. -
OpenAI Strawberry Project Leaks: Reuters Reports Internal Investigation Details
By
–
NUEVAS FILTRACIONES de OPENAI Un artículo de Reuters filtra información de una investigación interna llamado Strawberry, anteriormente conocido como Q* -chann channn!- El artículo la verdad que pone palabras a mucho de lo que se ha venido especulando en el último año: →
-
AI Vision Models Show Critical Limitations in Visual Understanding
By
–
Curious about the details? This research by Auburn University and the University of Alberta dives deep into AI's visual understanding limitations. Read the full study here, https://
vlmsareblind.github.io, to learn more! #AIResearch #TechInsights #Tech #Futurepedia -
Building Robust Enterprise AI Solutions: LLM Performance and Safety
By
–
At Cohere, we understand the importance of successful LLM implementation in enterprise settings. By learning the economics and performance of models, product owners can build cutting-edge, responsible AI solutions with significant value.
— Cohere (@cohere) 12 juillet 2024
Learn more: https://t.co/S8lTE14b0H pic.twitter.com/eKnxSbLDQEAt Cohere, we understand the importance of successful LLM implementation in enterprise settings. By learning the economics and performance of models, product owners can build cutting-edge, responsible AI solutions with significant value. Learn more: https://
cohere.com/blog/building-
robust-enterprise-ai-solutions-insights-on-llm-performance-safety-and-future-trends
… -
Tesla’s Self-Driving Program Uses Public as Crash Test Dummies
By
–
I thought this was going to be a story about ensuring that Tesla's #FakeSelfDriving behaved like an asshole driver, but in fact it's a story about how Tesla is using us all as crash-test dummies by relying on random YouTubers to test their bad software. https://
businessinsider.com/tesla-prioriti
zes-musk-vip-data-self-driving-2024-7
…