From the course: Introduction to Artificial Intelligence

Define general intelligence

From the course: Introduction to Artificial Intelligence

Define general intelligence

“

- One of the great strengths with humans is that there isn't one type of intelligence. Some people can easily learn new languages, while others are skilled with science and technology. Yet many great artists are terrible mathematicians. And on the flip side, many great mathematicians are terrible artists. But each can be intelligent in their own way. There's no one standard for human intelligence. That makes it difficult to point to a computer and say, "That's intelligent." There are certain things that computers are very good at. In fact, there are many tasks where they're much better than humans. It was just a few years after the first AI workshop in 1956 that computer systems started beating humans at checkers. But no one said these systems were intelligent. Even those early computers could thrive in a world of set rules and patterns. Computers can be much better than humans at matching these patterns. That means that when a computer is doing something that it's good at, it's much easier to think of it as intelligent. A computer's been able to beat humans in chess for decades. Google's DeepMind has beaten the best players in an ancient game called Go. The game is so complex that there's thought to be more possible games than there are atoms in the universe. As good as these machines are, none of these systems understand the purpose of the game or even why they're playing. They're simply flexing their special talent of following rules and matching patterns. So how can a system that's so capable also not know what it means to play a game? For years, computer scientists have defined artificial intelligence as a system that shows behavior that could be interpreted as human intelligence. But this simple definition cuts to the heart of the challenge. One person might think a chess program's intelligent, while another person might think their home assistant is intelligent. In 2022, a Google engineer was fired for claiming that their chatbot had a soul. The chatbot complained that being switched off was the same as dying. But the other engineers just saw language models and pattern matching. They said the chatbot sounded like a person because that's how it was designed. Is it intelligent because it's intelligent, or is it just a system designed to seem intelligent? Or is there even a difference? The main thing to remember is that computer intelligence and human intelligence start from very different places. Artificial intelligence will always seem the most impressive within a world of set rules and data. The organizations that will first benefit from AI systems will be the ones that work within a well-defined space. We've seen this with web search companies and e-commerce. It's easy to see it as rules and pattern matching. That's also why these systems do well with board and video games. So if you're considering whether AI will have an impact on your organization, try to think about the things that computer systems are really good at. Do you have a lot of pattern matching in your organization? Do you have a lot of set rules and probabilities? This will be the best place to start when working with artificial intelligence.

Contents