What people thought AI would do vs what it’s actually doing

A viral Reddit thread titled "What people thought AI would do vs what it’s actually doing" has struck a nerve with tech professionals, crystallizing a widespread feeling of unmet expectations. The discussion, sparked by user PrestigiousPear8223, challenges the dominant narrative that AI would rapidly automate work and grant unprecedented freedom. Instead, contributors describe a reality where tools like OpenAI's GPT-4, Microsoft's Copilot, and Midjourney serve as powerful but imperfect assistants—great for brainstorming, drafting, or generating code snippets, but far from the autonomous agents that handle complex, end-to-end job functions. The sentiment points to a 'reality gap' where the hype of artificial general intelligence (AGI) collides with the practical, often messy integration of narrow AI into existing workflows.

This gap reveals a critical phase in the technology adoption cycle. Professionals report that AI has changed their jobs, but not in the revolutionary way many anticipated. For developers, GitHub Copilot suggests lines of code but requires expert oversight. For writers, ChatGPT helps overcome blank-page syndrome but lacks nuanced understanding. The tools provide a 10-30% productivity boost on specific tasks rather than a 100% reduction in workload. The thread's popularity suggests a collective reassessment: the promise was a robot that does your job, but the current reality is a sophisticated tool that changes how you do your job, demanding new skills in prompt engineering and AI oversight without eliminating the core cognitive load.