Amazon tested warehouse robots and found they're not ready to replace humans

> "A key challenge of deploying learned VMPs is their lack of interpretability in failure cases," Amazon’s researchers observe in the Pick robot paper. "While it is often easy to fix a simple bug or improve algorithmic limitations of heuristic approaches, VMPs have to be re-trained or fine-tuned to learn about failures while maintaining their previous performance."

Exactly! This has been the issue with ML and especially NN based approaches for a long time. Really disappointing that we still haven't solved this problem. It's a real limitation.