Narrow AI vs General AI

Every AI system you have ever used is narrow AI. It does one thing, or a small cluster of related things, and it does them within tightly defined boundaries. Your spam filter cannot drive a car. Your image recognition system cannot write poetry. Even the most impressive large language models, which seem to handle an extraordinary range of tasks, are still operating within the bounds of pattern-matching across text - they have no understanding of the physical world, no ability to set their own goals, no capacity for genuine reasoning beyond what their training data supports. General AI - sometimes called artificial general intelligence or AGI - refers to a hypothetical system that could match or exceed human cognitive ability across all domains. It would learn new tasks as flexibly as a person, transfer knowledge freely between domains, and adapt to genuinely novel situations. Nobody has built one. Serious researchers disagree about whether it is decades away, centuries away, or fundamentally impossible. For practical business purposes, the distinction matters because it helps you calibrate your expectations. The AI you can buy today is narrow, and planning around it means identifying specific tasks where it can add value rather than expecting a general-purpose thinking machine.