@arish
Arish Ali
@arish · 1:46

Can AI develop intuitive physics understanding? Asking Dr. Ronjon Nag, AI pioneer

article image placeholderUploaded by @arish
So hi Ranjan, this is question for you. In the context of the work being done by the project, and I'm sure there are other initiatives as well. Can you explain if it is possible for Ali system to build this kind of common sense understanding of the real world, of intuitive physics, of what's going on? Or is there something fundamentally different in how humans develop this understanding and how current AI models do? That will make it really hard. Thank you. Bye

#TheCuriousMind #TalkToArish #ai #physics #r42institute https://s.swell.life/STBgdwKVeFpSpL3

@ronjonnag
ronjon nag
@ronjonnag · 1:58
But also, I think looking at physics purely from the data, it might be better to think about what could be actually calculated from the mechanisms of fundamental physics. So could we actually just use regular physics like Isaac Newton did and Albert Einstein, as opposed to looking at every single data point to get a more accurate formula? So thank you very much. I'll stop there as a few viewpoints
@arish
Arish Ali
@arish · 0:16

Graph Neural Network https://s.swell.life/STCtfvJI1miO2J8

Hi Ronjon, thank you so much for your response. Graph neural networks sound really cool. And for anybody who is interested in learning more, I've added a little YouTube link here to check out Graph networks similar waiting for the extent action
article image placeholderHow DeepMind learns physics simulators with Graph Networks (w/ author interview)
0:00
0:00