I have a collection of k-pop videos that were converted to 3d with AI and while there's loads of odd artifacts it can at times also look impressive (some content works better than others). That said these were converted in many hours of processing per minute of video. Also amusing is when people are using screens (phones, televisions) in those AI conversions they are also 3d, sadly in a way that current display tech can't do without sacrifices. There are some people in this thread saying 3d is pointless and lacking in consumer demand but I think it's largely just a technical problem with how the display tech has worked and how the content is produced. It would just be normal if you could do it without sacrifices.
Previous 3d Blu-Ray conversions don't have visual anomalies like AI conversions but they required a lot of work and often looked like they only had a few layers of depth just like bad AI conversion but I think this will incrementally get better with AI that can segment objects, infer a depth map, adjust for motion, fill in missing data, etc. And what takes hours of processing now will likely also be possible in real time with better models. It's a slow incremental march.