The field responds

The field didn't debate what had happened. It just moved.

Within a year, deep learning entries dominated ImageNet. The hand-engineered approaches that had been the standard for decades disappeared from competition. Researchers who had spent careers refining them started retraining as deep learning researchers. Not a gradual shift. A pivot.

Error rates on ImageNet continued to fall, year after year, until they dropped below what humans could achieve on the same benchmark.

The techniques didn't stay in image recognition. Speech recognition. Natural language processing. Drug discovery. Medical imaging. Within 18 months of AlexNet, every major AI lab had reorganized around deep learning.

The approach that had been a minority interest, kept alive by a small group of researchers through two decades of minimal funding, had just become the only game in town.