Are # Copilot and # Gemini smarter than humans in coding? Not necessarily. # LLMs follow patterns from their training data based on the real coding patterns of
The author questions whether AI coding assistants like Copilot and Gemini are truly smarter than human programmers. They argue that LLMs merely replicate patterns from their training data, which often reflects the average, rather than exceptional, coding practices of humans. The author illustrates this by describing a debate with Copilot and Gemini about a UI design approach for Project AEye, where the AI initially struggled to grasp the author's alternative method. AI
IMPACT Questions the perceived intelligence of AI coding assistants, suggesting human programmers may still hold an edge in novel problem-solving.