A Watermark for LLMs – a watermarking framework for proprietary language models. 4 of 11
SAFETY
-
A masterclass on the dark sides of AI models
By
–
New video on Sunday at 6pm. We'll discuss one of the darkest sides of AI models. I'm very proud of this video, for its content, its form, and its message to our society. It is clearly already part of the channel's masterclasses.
-
Responsible AI Article Series and Upcoming Robotics Research Coverage
By
–
The latest article in series is Responsible AI. More articles that highlights other areas to come – In particular, excited for robotic one given recent excellent works from Google researchers like Robotic Transformer(RT-1) and SayCan.
-
ChatGPT servers wasted on attempts to make it racist
By
–
What society would look like if half of ChatGPT’s servers weren't eaten up by people trying to trick it into being racist.
-
AI Ethics: Accountability and Transparency in Data-Driven Systems
By
–
With increased use of AI, there is significant concern around ethics, bias, transparency, & accountability of technologically-driven systems.
— IndiaAI (@OfficialINDIAai) 28 janvier 2023
The key is to adopt practices that promote data-centric innovations with accountability & transparency. pic.twitter.com/jREUj4NQ1dWith increased use of AI, there is significant concern around ethics, bias, transparency, & accountability of technologically-driven systems. The key is to adopt practices that promote data-centric innovations with accountability & transparency.
-
NIST AI Risk Framework: Security Illusion Without Data Science Processes
By
–
Will the new #NIST AI Risk Management Framework foster a false sense of security? Without data science processes and platforms for continuous improvement and shared best practices, Domino's Kjell Carlsson fears it will. @VentureBeat @sharongoldman
: -
Responsible AI Discussion: Interpretability and H2O Products
By
–
Join @bhutanisanyam1 for a thought-provoking discussion on #ResponsibleAI with @ledell
, @_dynamic24_ + @pandeyparul
. We'll delve into the current states of #interpretability, using models responsibly, and how @h2oai products support these efforts. http://
ow.ly/ql3x50MBCNT -
ChatGPT as Full Self-Driving for Content Creation
By
–
#ChatGPT is the #FSD of content creation. Fun but use with extreme caution
-
AI-Generated Content Indistinguishable From Human Work
By
–
Un poco siguiendo la tendencia de lo que está pasando actualmente con las imágenes. Para no expertos, la mayoría de imágenes o textos que genera hoy la IA son "lo suficientemente buenos" como para no saber diferenciarlo de una obra original humana.
-
Synthetic voices improving toward imperceptible emotional authenticity
By
–
Siempre un gustazo contar con la opinión de expertos! Lo que creo que aunque a día de hoy aún las voces sintéticas no están a la altura, en lo técnico seguirán mejorando hasta el punto en el que no tengo claro si el usuario final sabrá diferenciar si está ahí o no ese sentimiento