Caring for humanity is different from caring for humans, in the same way as global maximum is different from local maxima.
@aravsrinivas
-
Preserving consciousness in the age of artificial intelligence
By
–
Preserve the light of consciousness
-
Larger proprietary models outperform open source alternatives
By
–
Ya, that's true. Plus larger models than the OSS models.
-
Tools and Methods for Exploring Different AI Models
By
–
What do you currently use for exploring different AI models?
-
Adversarial Examples Research: From Neural Net Skepticism
By
–
Remember when writing papers on adversarial examples to say neural nets don't work was the trend at one point, not too long ago
-
Bullshitting Indicator: False Claims of Potential
By
–
"I could have been X if I wanted to" is probably the number one indicator of a bull-shitter.
-
Perplexity.ai addresses scaling challenges and potential pitfalls
By
–
It does to some extent, and we try our best to not let http://
perplexity.ai fall into this trap. But with full awareness it could happen as we scale. -
Perfect End to End Robot Demo Showcase
By
–
The perfect end to end robot demo doesn’t exis…. pic.twitter.com/RocsdIOmo6
— Aravind Srinivas (@AravSrinivas) 24 septembre 2023The perfect end to end robot demo doesn’t exis….
-
Decouple Personal Failure from Idea Failure Learning
By
–
When something you tried fail, it’s good to decouple your individual failure and the idea’s failure. It’s self-serving to write the failure off as the idea’s failure and go on to do the next thing. You will likely continue to fail if you didn’t learn your individual lessons.