AI-complete

AI-complete, also known as AI-hard, refers to problems in artificial intelligence that are considered to require human-level intelligence to solve and are as complex as achieving strong AI itself. These problems typically involve tasks that are easy for humans but challenging for computers, encompassing understanding natural language, recognizing objects in images as humans do, and exhibiting common sense reasoning.

The term implies that solving such a problem is equivalent to creating a system with general intelligence across a wide range of tasks, not just specialized intelligence for specific tasks. AI-complete problems are considered benchmarks for AI research, indicating progress towards creating machines that can truly understand and interact with the world in a human-like manner.

‍

Natural language understanding (NLU) is often cited as an AI-complete problem because fully understanding human language requires not only parsing syntax but also interpreting context, humor, irony, and the speaker's intentions, which are capabilities associated with strong AI.

Another example is common sense reasoning, which involves making the kind of everyday judgments that humans find trivial, such as understanding that when someone sells a car, they no longer own it, or that water will spill if a glass is turned upside down. These tasks require a breadth and depth of world knowledge and the ability to apply it flexibly across different situations, a hallmark of human intelligence that remains a challenge for AI systems.

‍