Artificial Intelligence in Corrections

Reading about the accelerating evolution on Artificial intelligence (AI) capabilities, I’m convinced this technology has a huge potential to disrupt a lot of business and even our society. AI could make business radically more efficient and answer questions we didn’t even know we had. But it may also destroy millions of jobs as machines could get better than humans at everything from driving cars, handle the most complicated customer requests to analyzing medical scans. It is difficult to predict how this technology will change our world.

Artificial intelligence is an area of research into whether a machine, such as a computer or robot, can execute tasks typically requiring human-like intelligence. AI researchers test the machine’s possibilities in key areas such as learning, reasoning, problem-solving, context perception and language. Today’s machines can already perform a whole range of impressive things. It’s amazing to see how IBM’s Watson wins the Jeopardy TV quiz or learn on Youtube about Google’s Deepmind recent breakthroughs in training deep neural networks.

As machine learning and AI is already improving medical diagnosis and decide on tailored treatments, is slowly entering the world of mental Health and Psychology and we see the first experiments with Prison Guard Robots, we can assume that this kind of technology will enter the correctional world sooner then we’ve expected. But before becoming frightened they will take over, we need to realize that machines aren’t that good yet. Machines are developed with a specific functions and skills and they work best when they are implemented based on what they are made for. Machines can become very good at one particular thing but they are far from good at multi-tasking.

But as machines can already be built to do particular things better than humans, I believe it would make sense to investigate potential usage in Corrections for those things we are not very good at. Maybe one example is risk assessment. I’m not at all an expert on this, but it seems to me that current assessment models are mainly based on a lot of valuable intensive research. I’m wondering what could be reached if those models could be permanently evaluated and improved by using continuous evaluations, gathering data for all offenders and using machine learning technologies to improve them.

As artificial Intelligence is by some understand as the attempt to artificially create cognitive beings, current existing AI technologies have fallen far short of making the predictions of 'having thinking machines within a few decades' come true. However the capabilities of those technologies are impressive and in my opinion it would be definitely interesting looking into their potential usage in the corrections domain….