[ Computers ]

AI Predictions

Do you find it convenient or creepy when AI makes a suggestion? Imagine that you are searching for a funny video, but you don’t know what it’s called or how to describe it. All of a sudden, the search bar goes: ”Oh, I know what you’re talking about – it’s the one where the baby keeps giggling at a cat.” After a few videos like that, you decide that you want to show a new, but similar, clip to your family, only to find it in your recommended videos. It’s handy, it’s scary, and it goes even further.

Predicting Behavior

How would you feel if I told you that AI was capable of predicting how you would act in a certain situation? The research that is done in the development of self-driving cars or predicting the outcome of any human interaction. AI is capable of using its deep learning to figure out how you are likely to behave.

When you think about it, there is a dose of reassurance in that. AI will not know what you’re about to do 100% of the time. We still have free will and, contrary to parameters of many experiments and algorithms, human beings do not always act rationally or in their best interest. However, based on the average actions in the past, it is able to predict what you are considering as your course of action.

Let us consider YouTube videos for a second. If you, like me, enjoy stand-up comedians, you are likely to be drawn to a certain niche. You might like comedians that use saucy language, criticize a certain political party, or offer their insights on humanity as a whole. Your feed will start featuring comedians that are the part of the same niche since previous users who watched George Carlin also watched Bill Hicks.

Deep Learning

To effectively predict what you (want) to do, AI uses something called deep learning. It is a method of learning that can be supervised or unsupervised. The same technology is used in facial and audio recognition, computer vision, machine translation, and many other areas where machines perform better than human experts.

This is why Facebook, for example, is able to recommend tagging a particular friend based on the image you uploaded, even if this friend is not connected to you via the platform. Facial recognition and the usual action of tagging people is now improved.

What Does This Mean?

AI is capable of learning on a level that can circumvent faulty input by humans. In other words, machines are becoming more capable of coming up with outcomes of certain dilemmas we would need some time to solve. Commerce thrives because of what the system predicts you want to buy.