Why AI Will Always Need Human Oversight, No Matter How Intelligent It Gets

The vast majority of tasks these days are automated through the use of a machine learning algorithm, and a big part of the reason why that is the case has to do with the fact that there is a routine that can be followed and based on this routine the artificial intelligence (AI) can end up ascertaining how much inventory is needed, assess customer satisfaction as well as do all sorts of things that companies rely on in order to function in the best way possible.

If you have been depending on these machine learning algorithms for quite some time now, you might just be assuming that things will always work perfectly no matter what happens because of the fact that nothing has gone wrong yet. With all of that having been said and out of the way, the Covid-19 pandemic has proven that this might not be all that true. The pandemic has changed everything from buying behavior to supply and demand and these changes are making it impossible for algorithms to function as efficiently as they used to.

According to Will Douglas, author of the MIT Technology Review, this is a good example to describe the importance of human oversight when it comes to AI. No matter how sophisticated an AI may be, at the end of the day it is only as good as the data it is being fed. While the purpose of machine learning is to adapt to changes, the truth of the situation is that whenever a change occurs this quickly it can be more or less impossible to adapt to for the AI in question, and this is why we are seeing algorithms and machine learning models become so efficient in the crisis that we are currently going through.

Jason Phippen who works at SUSE as the global head of product and solutions also brought attention to the fact that a change in the data stream could lead to the algorithm behaving oddly, and this could be particularly dangerous due to the reason that it would have an impact on the kind of decisions that a business ends up making. One can only imagine how catastrophic it would be if a business starting using tactics that were based on AI recommendations only to find that the recommendations came from an inability to deal with a massive change in the type of data that was being seen.

Human oversight in this regard could be as simple as data management. We are not quite able to enjoy self sustaining systems just yet, so at the very least managing the data that goes into the machine learning algorithm can make it so that sudden changes can be eased into the AI framework, allowing it to adjust and thereby ensuring at least to a certain extent that the decisions that are being made are going to be reasonable in some way, shape or form.

This is very important when you consider just how much more we are going to start relying on data streams in the future, and how AI and machine learning algorithms are going to end up playing an even larger role in our lives in the coming years.




Read next: An Artificial Intelligence Researcher Has Developed An Algorithm That Transforms Humans Into Animorphs

No comments:

Post a Comment