If you are wondering how advanced artificial intelligence (AI) systems are becoming, this news will clear the picture. The US military is testing an experimental AI network to identify likely future events worthy of greater attention days before they occur. Does a little "Minority Report", doesn't it?
The test program is called Global Information Dominance Experiments (GIDE) and was presented at a press conference last week. Combine data from un huge variety of sources, including satellite imagery, intelligence reports, field sensors, radar and more. To do what? Predict the future. Close, sure, but still future.
Foresight technology
Cloud computing also plays an important role in this setup, ensuring that vast chunks of data collected from around the world can be efficiently processed and thus accessible by any military officer and agency that needs it.
The GIDE project embodies a crucial change in how we use information and data to increase the decision-making space for leaders from the tactical level to the strategic level. It offers opportunities for not only military leaders to predict events, but also for our civilian leaders.
Glen D. VanHerck, U.S. Air Force general
The idea is to predict and anticipate other nations' moves in advance. A way to put deterrents and precautions in place before hostilities escalate or lead to something worse.
Planning events in advance? Some practical examples
An artificial intelligence capable of predicting events? It sounds like a lot, but I make you think about a few things. An example: if preparations are being made for a submarine to leave port, for example, it is quite easy to predict that it is about to set sail. What theartificial intelligence it really helps to do is use a lot more data, and a lot faster than any human operator can.
Another example: the number of cars in the parking lot of a hospital, or a military base, or a research station. If the AI sees an increase in activity, it can report this to other parts of the system, where this factor is then analyzed as part of a huge data set.
“The data exists,” says VanHerck succinctly. “What we're doing to predict events is making that data available, putting it in a cloud where machine learning and artificial intelligence look at it. Machines process them very quickly and provide them to decision makers, which I call decision superiority.”
This gives us days of advanced alertness and responsiveness. What in the past could have escaped us or took weeks of analysis, we are now doing it in minutes or almost in real time.
Glen D. VanHerck, U.S. Air Force general
Maximum confidentiality
Understandably, the US isn't revealing much about exactly how these new AI systems work to predict events. They are also quite buttoned up about how these systems process the information they are gathering. The final result, however, is imaginable: many, many more data processed in a faster time. GIDE has recently completed a third series of tests, and a fourth is expected shortly.
As mentioned, although at the moment the experiments seem a bit like Minority Report (people arrested for crimes before they have been committed) officials frame them differently. For them it's more of a form of supercharged information gathering, rather than a way to actually predict the future. To me it is somehow both.
Human beings at a certain level of power already make decisions based on data produced by mathematical models. Sometimes these are also models based on machine learning. AI will likely end up reducing choice options, rather than increasing them.
“The ability to see days in advance creates decision-making space,” says VanHerck.
