Agreed. Especially hallucinations are insanely high.
SAFETY
-
Inflection Joins NIST AI Safety Institute Consortium
By
–
We welcome NIST’s AI Safety Institute. Safety sits at the heart of our mission and culture at Inflection. We’re excited to share our expertise as part of the AI Safety Institute Consortium to help ensure that AI benefits everyone. #AISIC
→ View original post on X — @inflectionai, 2024-02-08 20:56 UTC
-
Understanding US Future: Technology and Societal Concerns
By
–
This thread helps explain my terror at the current/future state of the US
-
NIST AI Safety Institute Consortium Promotes Trustworthy AI Development
By
–
The AI Safety Institute Consortium established by @NIST brings together various AI stakeholders to support the development and deployment of safe, secure, and trustworthy AI. Here’s a statement from @StanfordCRFM and @Stanford RegLab leaders: https://
nist.gov/artificial-int
elligence/artificial-intelligence-safety-institute/aisic-member-perspectives#S
… -
Biden-Harris Administration Launches AI Safety Consortium
By
–
Just in: Together with @StanfordCRFM and @Stanford RegLab, we are pleased to join a consortium of 200+ AI leaders in academia, government, industry, and civil society dedicated to advancing AI safety. Read the news here: https://
commerce.gov/news/press-rel
eases/2024/02/biden-harris-administration-announces-first-ever-consortium-dedicated
… -
TGV and Shinkansen Safety Records Comparison Analysis
By
–
Have a look at the safety records of TGV and Shinkansen https://
en.wikipedia.org/wiki/List_of_T
GV_accidents
… https://
japan.go.jp/tomodachi/2014
/autumn2014/the_shinkansen_japans_high-speed_rail.html#:~:text=Japantakesprideinthe,tomaintainthisunblemishedrecord
…. -
Responsible AI Must Prioritize Environmental Protection First
By
–
Where companies also know the CSRD/SEC or other environmental regulations won’t come into full effect for 3-4 years they’ll use as much water as they want knowing at best they’ll get a rounding error fine. We must define responsible AI as showing the planet is protected FIRST.
-
Discussion on C2PA Metadata and AI Content Labeling
By
–
The watermarks – I assume should be the same c2pa metadata which can be removed from the image. As those images might get labelled as AI generated by major platforms
-
Critique of spontaneous behavior explanation as rudeness threshold tuning
By
–
I don’t buy that this is “spontaneous” behavior not seen in training. A simpler explanation is it was tuned to refuse some extremely rude requests, but it isn’t sure what that threshold for rudeness is.
-
Legal Liability of Healthcare AI Tools: Risk Mitigation
By
–
New policy brief: Who will be held responsible when the use of AI tools contributes to patient injury? Our latest brief explores the legal liability risks of healthcare AI tools and proposes risk mitigation measures. @MichelleM_Mello @NeelGuha https://
stanford.io/42tQsCp