The headline on a Washington Post article the other day read "Work computer use alerted Coast Guard to lieutenant's terror plans, prosecutors say". The story then started, "Christopher P. Hasson used his computer at Coast Guard headquarters to study mass shooters, according to investigators. An internal program, they said, picked up his suspicious computer activity."

To a large extent, this "internal program" is what we call artificial intelligence. Computers are used to scan information on other computers to find something "interesting". While this capability greatly expands human capability in many ways, I find it a misnomer to refer to this as intelligence. The computer, with these programs, is giving us capabilities well beyond what is humanly possible. The above story is an example where the computer was used to scan everyone's email and look for something suspicious. That is data gathering and filtering. A human must look at those results and decide yes or no on importance. That is intelligence. The difference is subtle but important. We do not yet have computer programs that can or programs we actually want, to make the final decision. The line from the astronaut in Hidden Figures really struck me as right on. There was a discussion as to whether to launch or not and the engineers were saying that the computers were saying it was OK. The astronaut wanted to know what the lady had to say.

As a scientist, I do not want to diminish the importance of what we call artificial intelligence. It is giving us capabilities beyond what was possible before. Computers can now be used to examine extremely large data sets and present information to scientists that allow discoveries that were not possible before. However, you still need the scientist to make the discovery.

Let me give an example of a possibility that relates to my field of research. Let's say there is a storm on the Sun, solar flare. The question is, what is the impact of this storm on Earth systems? We could have an "artificial intelligence" program that goes out and searches all the information on all computers throughout the world, everything that happened on Earth, during the next few days. The computer could then give the results of any anomalies like there were 10 times more automobile accidents during that time. And, there were 10 times more power outages. The computer might say that solar storms cause automobile accidents. The scientist might do some more research on none solar storm power outages and find that the power outage is what causes the increase in accidents. So, you need the scientist to do the final analysis that says solar storms cause power outages, and those outages lead to accidents. This type of artificial intelligence is a great tool. It allows humans to be more intelligent, but it is not intelligence by itself.

There is another type of artificial intelligence being tested at George Mason University that may really be artificial intelligence. They are using robots to deliver food to individuals. These robots move around campus like a person. They "decide" to take the handicap ramps instead of the stairs. They move out of the way of people. They stop at crosswalks and wait for a person to cross and then follow that person. I hear that one even made a mistake and got hit by a car. I'm not sure exactly how they are controlled, but it may really be built in artificial intelligence.

Artificial Intelligence does not yet have intelligence as we know it. What it has is expanded human capability. What it has is expanded data gathering and filtering and the ability to work with data sets that are beyond human capability. We are not yet at the point where we want to give final decision making to a computer program. Most of us would much rather see a human being making the final decision on issues that matter.