What Makes A Machine Intelligent?

Artificial intelligence is constantly improving, but will it ever be sentient? Will AI ever pass human intelligence?

The annals of science fiction are packed with stories of intelligent machines, computers that achieve sentience and go forth to either assist or menace humankind.

Modern computers are plenty smart, at certain tasks, but at what point can we truly call them intelligent? It all depends on how you define your terms. Julian Huguet tackles the big questions in today's DNews report.

The most famous assessment of artificial intelligence is the Turing Test, created by hero scientist Alan Turing all the way back in 1950. The gist: If a computer can converse with a human - in a text-based natural language conversion - and fool a judge into thinking it's human for at least part of the time, it's deemed intelligent. There's quite a bit more to it all, and plenty of competing interpretations of the test.

RELATED: Young Artificial Intelligence to Get Tech Nannies

Most researchers working in the field today agree that true artificial intelligence, in the science fiction sense, would need to pass more than just the Turing Test. To truly mimic the human brain, an artificial intelligence needs to do a lot of different things. For instance, A.I. must be able to "see" and interpret images. Recent advances in facial recognition suggest this is one area where computers are learning fast.

Other kinds of supercomputers are good at other things, like playing complex games. Earlier this year, Google developed an A.I. that could beat master-level human players at the ancient Chinese game of Go. But just because a machine that can do one task at or above human level doesn't make it intelligent.

Probably the single busiest area in A.I. research these days concerns the concept of machine learning. This is a method of designing computers so that they can essentially teach themselves. Machine learning systems use powerful algorithms to write their own code, refining it over and over, very quickly and virtually endlessly.

Given infinite data storage and processing power, this could potentially lead to computers that can teach themselves, well, everything. Check out Julian's report for more details, or click on over to our Seeker Stories special report: Is A.I. the Next Phase of Human Evolution?

-- Glenn McDonald

Learn More:

Washington Post: A computer just passed the Turing Test in a landmark trial

Gizmodo: The AI Revolution: How Far Away Are Our Robot Overlords?

NPR: Facebook's Facial Recognition Software Is Different From the FBI's.