All the schools are shut and the students are at their home in the wake of COVID-19 outbreak. The schools around the world are quickly shifting to online classes – this massive digital shift is a need of the hour but has some obvious and some other concealed and inconspicuous concerns.
While the obvious issues are the availability of the internet, whether students would easily be able to use the technology, teachers can get on to the speed and put the material online etc., the others are around the security and privacy.
We analyze the online classes and concerns in conducting the same because the education industry is going to change forever post Covid-19 and is also among the hardest hit industries.
The student data, generated through this shift to online classes, is sensitive and pose serious threats if compromised and used by criminals.
Data which we see as sensitive are:
- Videos containing students’ and teachers’ biometrics i.e. personally identifiable information such as faces.
- The geographical location
- Classroom activities
- Content of the online class
With these details, the miscreants can, in turn, get the social security number, Aadhar, Home Address, PAN etc.
There have been cases in recent past amid the current lockdown confirming these threats. One such case happened in an online class, using a popular mobile application, in a Bangalore school where non-students joined the meeting as the meeting ID was shared through WhatsApp messages and it reached the miscreants. There were unpleasant activities and the school stopped the online classes. This is easy to handle though.
How do you stop the identification of the people in the video?
Imagine all the students and teachers have their fake faces displayed in the video but the lip-sync, facial expressions and other movements are intact. The downside could be that the students can not identify their friends in the video may feel detached having all strangers in the class.
At the same time, if this is an online course where different people are expected, it would be fine.
The technologies behind these fake videos are from the recent advancements in Artificial Intelligence; they are mature enough to have good enough results. Computer Vision and Generative Adversarial Network (GAN), a special Deep Learning algorithm, are used for such applications. There are different levels of accuracy and quality achieved by different variants of the algorithms based on the training and tuning.
What about voice?
The voice modulation can also be done but is less sensitive biometric information.
Another issue would be the question-answer mechanism used by the teacher in a class. I have seen that teachers are struggling when it comes to question-answer and evaluation of what is being taught in an online class.
Text-based question-answer techniques are well researched and defined. They use Natual Language Processing (NLP) where the Deep Learning models (e.g. Google’s BERT, Flair and StandfordNLP) can be trained for the content being taught, questions can automatically be formed and handed over to the students.
The answers, which students submit, can also be automatically matched against the answers that are found using NLP algorithms using one of the data matching algorithms, allocating marks to the answers provided by the students.
It seems that most functions of a class can be automated and the teacher can just focus on explaining the concepts, to the students, in the best possible way.
Deep Learning, GAN, NLP and Machine Learning are the Artificial Intelligence (AI) technologies and help in solving the toughest problems emerging from constant digitization of the education industry or schools, attempting to match the experience of a traditional classroom teaching.
We, at DataHive Labs, expertise in solving end-to-end problems using artificial intelligence technologies. Contact us for any further discussion at [email protected]
