Dubai: Educational institutions in the UAE have started using artificial intelligence (Ai) and machine learning to predict students at risk of dropping out, the employability of graduates, and other patterns, according to a Dubai-based data science company.
In an interview, Raman Thiagarajan, founder of Nexquare, said schools and universities, faced with rising expectations of regulators and parents, will have no choice but to adopt the trend.
Nexquare, a start-up founded in 2015, is the ‘Global Winner’, selected from 6,000 international applicants, in the ‘Education Category’ for Dubai Future Accelerators (DFA) in 2017.
The DFA is a programme for “cutting-edge entrepreneurs, in partnership with the government of Dubai, to use the city as a living test bed for creating solutions to the global challenges of tomorrow”.
Nexquare is working with the Knowledge and Human Development Authority (KHDA), the education regulator in Dubai, and education institutions in using AI and machine learning to “re-imagine education delivery and management”.
Thiagarajan said the terms AI and machine learning are used somewhat interchangeably even though they are not identical.
However, both concepts involve computers that learn tasks independently, without being explicitly programmed for them by humans. An example would be using data about a student to say if the likelihood of the student dropping out will increase or decrease.
‘Pressing problems’
Nexquare has built such models for a college as well as a school group in the UAE, “plus a number of advanced analytics use cases for government entities to enhance the effectiveness of education delivery”.
Thiagarajan said he would not name the institutions or entities for reasons of confidentiality.
Speaking about AI and machine learning in education, he said the technology is being used “to address some of the pressing problems today in schools”.
Thiagarajan said machines can now predict if, for instance, a UAE student’s chances of dropping out are going to be higher or lower than average if the student chooses a certain course of university study.
Patterns within patterns
Machines are also looking for patterns within dropout rates, such as “are there some underlying drivers that are linked to economic factors, and are certain children migrating towards certain types of schools that is causing the dropouts. We are creating predictive patterns around those, using all the data that the school has”, he added.
“You could do that for teachers, see how likely a teacher will be successful at the school, or move on to another school.”
The machines use a mind-boggling variety of data — ranging from the student’s socioeconomic background, reported behavioural issues, scores, academics, attendance, assignments, curriculum, extra-curricular history — in discovering the patterns.
“A machine can fathom thousands of combinations in milliseconds, but also can understand the repetitive patterns and detect patterns that a human mind will not be able to detect … The model looks at all the variables and it knows which variables matter,” Thiagarajan said.
Predictive accuracy
When asked how accurate will be the predictions of the machines, which have only recently started working at educational institutions here, he said: “There is no easy answer to that. There are mechanisms to know the validity or accuracy of a predictive model, that are established in this field, in the science of analytics. But that is still theoretical.
“The reality on the ground — we will know that only with time, as it plays out. Schools will track this and see ‘did it make a difference, did it have an impact’. It will take at least a year, not months.”
Thiagarajan emphasised and clarified that the system is not meant to decide a student’s fate, but designed to be used as “a powerful aid” to assist counsellors and decision makers in flagging early warning signs and using appropriate interventions.
Higher responsibilities
He said schools and universities must understand that the predictive and analytical models rely on data, which has to be captured or recorded digitally — for a system and in a system specifically designed to make use of all that data in a meaningful, easy manner.
“Educators can no longer think of this as a hobby or a business, the responsibility is a lot more higher now … The KHDA — a forward-looking regulator — when they go to a school today, they are looking at data — the school’s ability to demonstrate, with data-centric, evidence-based approach, that they are actually having an impact on the well-being of the whole school.
“So, this is already beginning to happen; it will probably become far more acute as we go forward. Schools will need to evolve and be more far more data-centric and evidence-based in how they are going to claim they are making an impact. It helps in government school inspections and in their credibility with parents.”
‘Greater opportunities’
Hind Al Mualla, chief of creativity, happiness and innovation at KHDA, said: “Dubai Future Accelerators allows us to explore new possibilities which can really change the way education is delivered today. By combining data science and advanced analytics, Nexquare is developing greater insights on what’s really working in our schools and what needs to improve. It is exciting to see these new ideas which can open greater opportunities for both educators and learners in the future.”
