Engage with the chat bot on a customer service portal and you are likely to know something is not quite right. Interact with an animatronic robot, and that missing piece becomes even more apparent. In both cases, the thing that bugs your subconscious is that the machine you are interacting with is not socially aware. It can’t be – at least not yet.
We have come a long way in the fields of artificial intelligence (AI) and deep learning. We have come so far, in fact, that we now have machines capable of doing all sorts of mind-boggling things. But we have yet to build a machine that can interact with a human being on a personal, social level. Researchers in Australia and Germany aim to change that.
Personality Types and Eye Movements
There are a lot of aspects to human biology that are unique among our species. One of them is eye movement. Research has revealed that measuring and analyzing eye movement can indicate a person’s personality type. The Australian and German researchers have focused their efforts on four in particular: neuroticism, extroversion, agreeableness, and conscientiousness.
They have developed an artificial intelligence (AI) system that uses advanced machine learning techniques to analyze eye movements. They say that a person’s eye movements reveal some of what that person is thinking or feeling at the time. Eye movement can reveal curiosity, sociability, and even a sense of morality.
When you stop and think about it, the findings are fascinating. Leaving out AI and machine learning for just one minute, humans have long believed that the eyes are the window to the soul. Looking into a person’s eyes reveals a lot about who that person is and what he or she is thinking and feeling in the moment.
What the Australian and German researchers have done proves something we have known instinctively all along. More importantly though, they have developed a system that could someday translate into machines capable of interacting with us on a more human level.
Crunching All the Data
So, how does it all work? We do not have all the details, but it would appear as though some common principles of big data, analytics, and signal processing are in play here. Let’s start with the big data and analytics components.
True AI and deep learning do not really exist in their purest form. Machines cannot truly learn on their own. They are still dependent on making use of the data fed into them. And in the case discussed in this article, it takes a tremendous amount of data to fuel the AI and deep learning systems. Researchers required a huge data set just to get their system up and running. Then the system had to continually collect additional data and analyze it in order to come up with usable results.
Signal processing comes into play during data analysis. As the eye-tracking system collects data, useful information is collected alongside plenty of noise, explains Rock West Solutions. Signal processing is utilized to get rid of that noise so that only valuable data is used for deep learning.
If AI and deep learning were not complicated enough, signal processing is a very complex science with a lot of moving parts. A company like Rock West Solutions put as much energy into developing signal processing systems as the Australian and German researchers are putting into their AI and deep learning systems.
As research continues, the goal is to eventually make machines more socially aware. Can the combination of AI and deep learning pull it off? We’ll have to wait and see.