Impact Lab


Subscribe Now to Our Free Email Newsletter
December 23rd, 2016 at 10:52 am

Will machine learning be used to solve social problems in the future?

a

The answer is, quite frankly, yes. Machine learning is currently used in some ways, including solving social issues. Algorithms have been designed that use is predicting what movie you will enjoy watching, or what you might like to buy from a particular retailer. But now, things need to be stepped up a notch before we are ready to solve the world’s social issues by using machine learning.

IL-Header-Communicating-with-the-Future

One thing that we do need be aware of in using machine learning is the potential for misuse, and we can best prevent this from happening. But, once that has been figured out, the potential for good things to come from using machine learning techniques is enormous. It could be applied to a dataset of bond court cases to decide the outcome for the defendant. Decisions such as whether they should be released or remain in jail as at risk of re-offending could be left to a computer through the help of an algorithm.

Statistics suggest that by using this type of prediction algorithm we could potentially reduce jail populations across the country by several hundred thousand people. These machine learning tools are much cheaper and easier to employ than programs and police staff that cost millions of dollars. In an application that deals with pre-trial decisions, the built-in algorithm searches for past defendants similar to the one before them and uses existing crime data of these like defendants as the basis for making its decision.

Machine learning algorithms are best used in situations where there is lots of previous data to work from, and the outcome is one that’s measurable. But one must be aware of the possibility that any prediction system could also be biased with the underlying data used based on discrimination. Any incorrect or misleading information could seriously alter the outcome of any machine learning tool. And, just because the algorithm is able to perform well during testing, that doesn’t mean it will do the same in real life situations.

There is still a lot that needs to be uncovered about machine learning that we just haven’t managed to do yet. One example being: how to combine the thought pattern of a human and that of an algorithmic judgment to get the best possible outcome. The makers also need to consider at what point the algorithm should be overridden. But, it’s a field that’s constantly being worked on and as a result, is advancing at a phenomenal rate, so watch this space as this is just the beginning for machine learning.

Image Credit / Article via trendintech.com

IL-Header-Communicating-with-the-Future

Comments are closed.

First User 4