There is real-time raytracing...

...but why there is still no strong AI? I mean two problems are related, since raytracing creates photorealistic image of the geometry, while AI has to decipher it correctly.


Yet navigating arbitrary real world environment is still an unsolved problem. How so?

Raytracing being "real-time" is merely the consequence of hardware optimization, RTX cores that are built for the task(AMD has something similar coming).