Can Someone Please Tell Me What AI Actually Is?

Since the discussion of AI began showing up on social media platforms, in the news, and in general day-to-day conversations, using the term has become second nature. I’m sure you would get a blank stare if you paused a conversation to ask what AI actually is.

The textbooks tell us it’s, “The ability of a machine to perform cognitive functions we associate with human minds, such as perceiving, reasoning, learning, and problem solving.”

This definition seems to suggest that computers and humans are similar in their information processing. The brain, in fact, acts nothing like a computer. It is a system that puts together patterns among neurons. Dr. Ralph Greenspan says:

Computers record, and computers have things stored in specific places that are stable. Our brains do none of that. We do pattern recognition. Even though we are capable of logic, our brain does not operate by the principles of logic. It operates by selection of pattern recognition. It’s a dynamic network. It’s not an “if-then” logic machine.

With this added insight we land on a more precise definition of what AI is. Artificial Intelligence is the process a machine undertakes to solve problems that we associate with human cognition including: perceiving, reasoning, learning, and problem solving. The key difference is the methods by which the computer solves these problems.

Humans and computers may take on similar problems and may come up with similar answers, but the process by which they go from question to answer is completely different. This also gives us a better idea of what AI is capable of. AI is very good at solving logical problems of a narrow scope. Games like chess are great for AI as there are set rules and a clear winner and loser. Problems that require improvisation, creativity, and have ambiguous outcomes prove much harder for computers to take on.

This difference leads to computers being better than humans at solving certain sets of problems, and being unable to offer any clear solutions to others. Check out some things AI has failed to tackle here.

Stephen Hawking and Elon Musk have warned that AI may eventually achieve the capacities needed to enslave the human race, but from what we know so far, as long as the problem of world domination remains an exercise of improvisation and creativity, us humans remain safe.