Deep learning has already proven its worth. Google translate is an example on the older side. As LLMs go, I can take a picture of a tree or insect and upload it

Deep learning has already proven its worth. Google translate is an example on the older side. As LLMs go, I can take a picture of a tree or insect and upload it and have an LLM identify it in seconds. I can paste a function that doesn't work into an LLM and it will usually identify the problems. These are truly remarkable steps forward. How can I account for the cynicism that's so common on HN? It's got to be a psychological mechanism.