What’s something AI does surprisingly badly that everyone assumes it should be good at by now?

A Reddit discussion in the dev_community asks what AI still performs surprisingly poorly at, despite common assumptions of its proficiency. The original poster clarifies they are not referring to well-known issues like hallucinations, but rather tasks where AI's continued inadequacy is perplexing, prompting users to question why certain functionalities remain underdeveloped.

This Reddit thread stands out by focusing on AI's persistent, less-discussed shortcomings in practical applications, unlike typical discussions that center on well-known issues like hallucinations.