[bcks]

Back in 2018, I ran a little test to see if I could push Google Cloud Vision to recognize objects, shapes, or patterns in clouds. No matter how I treated the images ahead of time, the answer always came back: clouds.

Would be interesting to see how much free-association and hallucination have "improved" the results with the current generation.

[a_t48]

Your problem was right there in the name - google "cloud" vision.

[4gotunameagain]

I think it highly depends on the technique.

The image recognition of google at 2018 most certainly was trained on a database of labelled images, and I would bet that the labels should have been short and distinct, not "a cloud with a vague representation of Bruce Willis' face" !