Artificial intelligence (AI) is not considered to be generally intelligent like humans, but it can perform specific tasks very well:
- Narrow intelligence: AI is currently considered "narrow" intelligence, meaning it can only perform a limited set of actions based on its programming. For example, an AI that can identify cats in images can't perform natural language processing.
- Human-like tasks: AI can outperform humans in some tasks, such as strategy games, language performance, and writing essays. However, AI can also make mistakes, or "hallucinate", just like humans.
- Lack of independent thought: AI is programmed by humans, so it can't think or reason independently. For example, an AI that can identify cats in images doesn't understand what a cat is.
- Potential for human-level intelligence: Some experts believe that AI could eventually achieve "human-level intelligence" (AGI), or proficiency in the many domains of human intelligence.
AI can be useful for automating dangerous tasks, such as handling explosives or working in deep ocean water. Self-driving cars and other vehicles also have the potential to reduce the risk of injury to passengers.