Could AI help students pick their CAO courses?

A UCC startup hopes to help students make the difficult choice by analysing past data

What CAO courses to put on their list is arguably the hardest decision students have to make in the course of their secondary education. At the time, the list of third-level courses seems to be fixing the course of their future, their career, their possible income, and even future happiness. Many factors need to be taken into account: the student’s aptitudes for different subjects, their interests, what universities are feasible for them to attend given location and financial factors, even the job prospects in various fields. The pressure placed on these 17-18-year-olds is immense.  

A new Cork-based start-up, Yooni, aims to use artificial intelligence to help students make these difficult decisions. The company was launched by two UCC electrical and electronic engineering graduates, Darragh Lucey and Nathan Mayes. The software began in conjunction with UCC and Cork Institute of Technology but is now expanding to include recommendations from third level institutes nationwide. It has already been used by students from over 100 schools, suggesting that, though this idea is very novel, there is a market for this sort of AI guidance. 

Yooni brands itself as “college course recommendation software” and claims to be able to provide students with their optimum college courses in 15 minutes. Firstly students input the subjects they are taking for leaving cert and a prediction of grades they will achieve in the June exams. The software then compares this data with that of current third level students. It examines the leaving cert subjects and grades of those third-level students, what college courses they went on to study, and their academic performance in those courses. It uses this information to assess which courses the second level student should academically excel at. 

Whilst the Yooni software includes personality tests and career aptitude tests, these are all separate, meaning that students could get three very different sets of results. These additional tests are also not carried out using the AI, data analysing element so in essence Yooni’s main results are still those predicting which a student will excel at academically. Yooni says, Our recommendations are based purely on your academics and do not take interests or other factors into account, meaning they are completely unbiased.”

This raises questions about whether these results should be unbiased. Do the elements that bias our results, our interests, and passions, and the things that get us up out of bed in the morning not count for as much, if not more than our academic aptitudes? On a personal note, I know if my Leaving Certificate grades were plugged into a software system it would predict I would currently be studying German. But due to placing more importance on the subjects that I loved, I’m three years into a Chemistry degree which, despite probably not being my optimum academic area, keeps me interested and engaged every day. Even on the days that I’m close to pulling my hair out over Schrödinger’s equation or something equally difficult, I never regret picking the course I love rather than something I would have found easier. So I find the notion of making decisions on “unbiased” academic ability a little problematic. 

In addition to this, presenting students with three different sets of courses based on their academic ability, personality, and career interests, could lead to further confusion and stress for students in making their decision. Yes, if a student’s tests align nicely this is a valuable confirmation for them that a certain CAO course could suit them well. But if on the other hand, a student’s course results are very conflicting this could add to their confusion. At the fairly steep one-time fee of €69 for the use of the software, students would hope to gain more clarity rather than greater confusion. 

Also in recent months, we have seen the potential for algorithms to fail at crucial moments for students. After the mistake in the code used to predict Leaving Cert grades this summer, at least 6,500 students were left with a lower grade than they deserved. An issue in a single line of code out of 50,000 had a major impact on students, and this is something that needs to be considered in the case of software like Yooni. The software needs to be thoroughly tested and examined when it is being used to make decisions with such gravity. 

A very positive aspect of Yooni is its Carpool Courses series, inspired by the Carpool Karaoke segment of the Late Late Show with James Corden,   where they informally interview graduates of various degrees. Particularly in an exceptional year like this, where on-campus open days have been forced to move online, it is important for secondary school leavers to get a real feel for courses they are interested in. It has to be said that the best way to find out what it is like to be a student of a particular course, is to talk to those very students. Informal, personal opinions and experiences of courses are a valuable resource to students.  

While the general idea of Yooni is interesting and certainly in keeping with current moves towards increased use of AI, it is a change that certainly won’t sit easily with everyone. You could argue that there are some decisions which machine learning shouldn’t be allowed to meddle in, while others may think that companies like Yooni enable students to be more informed about their compatibility for a course. But with the Higher Education Authority, reporting 1 in 4 students do not complete the college course they began, questions need to be asked about whether students are informed enough when choosing their courses, and AI-based or not this certainly needs to be dealt with.

Lucy Fitzsimmons

Lucy Fitzsimmons is the SciTech co-Editor of Trinity News, and a Junior Sophister student of Chemical Sciences.