Not everyone has experienced visual errors with existing models but it's pretty easy to find one. Just ask any frontier model to count more than 10 objects and they'll make mistakes. Last week, I went in depth with @FirasSozan on Inside the Silicon Mind to talk about the visual reasoning we're building at @ElorianAI.
Visual reasoning is a fundamentally different problem to object recognition.
No Digg Deeper questions have been answered for this story yet.
Not everyone has experienced visual errors with existing models but it's pretty easy to find one. Just ask any frontier model to count more than 10 objects and they'll make mistakes. Last week, I went in depth with @FirasSozan on Inside the Silicon Mind to talk about the visual reasoning we're building at @ElorianAI.
Visual reasoning is a fundamentally different problem to object recognition.
See the full video here: https://www.youtube.com/watch?v=CoaWmzkYFak
