AI Pushes Frontiers of Special Education


Over the last ten years, AI has caused a tremendous upheaval in how we do things from how we run our homes, conduct our business, and even how we “drive” self-driving cars. But it has also been quietly changing the face of special education and pushing frontiers that can integrate those with special needs better in work environments. AI tools are capable of mimicking supreme intelligence – and at the other end of the spectrum, enhance the manner in which children or adult students interact with their environment to promote learning.

“Nowadays, there are plenty of AI based applications tailored for people’s educational needs such as learning new languages or improving knowledge in different areas.” – Ilker Koksal, Forbes

The ultimate goal is to provide intervention strategies based on an accurate diagnosis. Children often show one type of disability and not the global range of impairments making “a magical cure-all” impossible to succeed. And this is where the beauty of AI-assisted intervention lies: it can be differentiated to target specific needs like language, reading, writing, and math difficulties.

Surprisingly, AI systems pretty much like how autistic people think. Teaching Robots, far from those we see in Pacific Rim and other sci-fi hits are not killing machines but gentle personalities that are used to socialize children with autism.

Universities like MIT, Vanderbilt University and Notre Dame backed by companies like Softbank are developing robots that teach children with intellectual disabilities to think, connect, and communicate. A huge part of the socially unacceptable behavior exhibited by these children stems from the fact that they are frustrated about being different and not understood. In an amazing turn, robots captured the imagination of autistic children so that they learn to imitate and understand human emotions. From preferring isolation, these children show more confidence, focus, and a strong desire to interact with these robots that are simple, predictable and not at all intimidating.

Currently, more inroads are being made in the areas of Autistic Spectrum Disorder (including Asperger’s) rather than for Downs Syndrome which is classified as mental retardation. Autistic children who used to struggle with verbal or non-verbal communication to the point where they misinterpret verbal and visual cues have improved using AI assisted technology. Sensory issues that bother these children have diminished and they can now tolerate being touched and were observed to maintain better eye contact. In fact, even if they are adverse to noisy environments, working with toys made them tolerate “talking robots” and were even seen to touch these curious-looking “toys” without prompting.

Some examples of robots that had successfully engaged children with special needs include Kasper, a learning assistant, developed by the University of Hertfordshire’s Adaptive Systems Research Group. Kasper, a social robot taught children how to feed themselves with a spoon and fork, comb their hair and brush their teeth. It was such a huge success that when Kasper was removed from the play environment, children with special needs continued to interact better with their playmates. Looks like Casper, the friendly ghost has been incarnated in Kasper, the friendly robot.

Milo, the robot developed by Robots4Autism, is best with children up to the intermediate years. This time, instead of working on hygiene and basic social skills, he levels up by helping special kids understand emotions so that they can respond in a socially-accepted manner. It’s a robot that understands how these children can tune-out sound by having a monitor on its chest that helps them focus. More importantly, just like Kasper, it has a gamut of sensors that include facial recognition to evaluate progress. It’s hefty $5000 includes pre-programmed curricula developed by experts which can be controlled by teachers from a tablet. The technology for using these teaching assistants in special education classrooms used to be a dream but it definitely has become a reality with Milo.

There is definitely more in store in the future. Already, Jain, a researcher has come up with a model for higher order learning called PLEDDOR or perceptron based learning disability detector. It is an artificial neural network that identifies reading, writing, and math difficulties bases on tests created by special educators. This article has barely touched the surface of what is available for smart-teaching. DiegoSan, Penguin, LIFEisGAME and more are on hand to improve the future for kids that use to be marginalized.