GENERATIVE AI

TTS Systems Show Improved Error Recovery Despite Occasional Gibberish

By

–

14 May 2024 6h48

TTS' still spews gibberish sometimes, but the realization and error recovery in the first example is astounding 😯 https://t.co/krMFGypACE
— Dr. Angelica Lim @petitegeek.bsky.social (@petitegeek) 14 mai 2024

TTS' still spews gibberish sometimes, but the realization and error recovery in the first example is astounding

→ View original post on X — @_jasonwei

GPT-4o Achieves State-of-the-Art on GPQA Benchmark

By

@_jasonwei

–

14 May 2024 6h33

Nice result lost in the hype: GPT-4o reclaims state-of-the-art on GPQA with 53.6%, overtaking Claude Opus (50.7%).

VLMs Emotional Knowledge: Beyond Simple Smile Detection

By

–

14 May 2024 6h13

Emotion recognition ≠ social signal detection – smiles don't always mean you are happy. But, my experience with VLMs suggests there's a lot of latent emotional knowledge and reasoning thanks to all the ingested novels. More testing and demos needed!

Multimodal AI Expert Reviews OpenAI Demo on Emotion

By

–

14 May 2024 6h13

I've been working in multimodal AI for human-robot interaction research field for 15 years, specializing in emotion and expressivity. The @OpenAI demo was freaking impressive. Some thoughts.

AI translation breaking language barriers: capabilities and risks

By

–

14 May 2024 6h13

The use of social AI to overcome language barriers has been an everlasting use case that I think everyone can get behind. It does something humans really can't: speak tons of languages! It's C-3PO! But translation needs to be careful. You won't know if it's wrong.