December 8, 2020
AI in education marketing and recruitment is a hot topic, and no question about it. AI is set to revolutionize much of our world in ways we can't yet quite be sure about, but even now there are advances that are nothing short of incredible.
We'll be doing a two part article on AI, with part one looking at its application to Education M&R, and part two looking at AI in education itself. First, let's clear up a common misunderstanding about which advances in our field are automation and which are AI.
Well let's begin by saying that knowing the line between automation and AI isn't actually that easy. In general terms, when a human gives a machine strict parameters and the machine always operates within those parameters, that is classic automation.
Examples of automation would be when someone clicks on the "subscribe" button on your landing page and automatically receives a welcome email with their name at the top. Programming your social media posts with HubSpot to go out at a certain time, and setting the analytics to measure engagement is another example. Automation can get a lot more complex, such as when the prospective student opens an email about a particular course, thereby showing interest, and is moved to a high priority workflow for more intensive follow up from the admissions team. This is still automation, however. The parameters and workflows are all predefined, and the software can't "learn" from the tasks it carries out. Automation is something that saves a huge amount of time, effort, and resources, and completely changes the results you can get. We know because we work with automation every day, and we see the substantial difference it makes to our clients and partners. AI on the other hand is much more complex. We have heard tales of an increasing number of companies claiming their product is AI, when it really is automation. While automation can be complex, it can't use the data it analyzes to make its own predictions, and that is the key difference. Even if the decision tree is multi-level, if it always follows set patterns, it is automation. If it learns from the decision tree to make new predictions and recommendations and then changes the processes in that decision tree, then this is machine learning, otherwise known as AI. Phew!
A study commissioned by Microsoft found that more than half of US Universities already use AI, and just over a third have AI as part of their core strategy. It is commonplace to find that AI was first introduced in the Marketing and Admissions department, but to do what?
Some US Universities use AI to mine data on prospective applicants; recording exactly how quickly they open the marketing email, and how long they spend reading it, so that they can prioritize leads by those "demonstrating interest". This data is then fed back into complex algorithms to determine the efficacy of the marketing strategy and to recommend changes for the next approach.
Personalized services can also come from AI-powered processes. For example, AI is being used now to identify "summer melt" applicants, who pay their deposits in May but don't show up in September. These applicants will receive personalized text messages, and other interventions to support them in following through on their initial decision.
The same technology can also help to identify students who may potentially engage more with their studies, or may in fact need some extra support; something we'll look at in part two of our AI feature next week.
Where things get seriously interesting, however, is in the area of student selection and admission.
This is where things start to feel pretty futuristic. Alice Gast at Imperial College of London is one of several key figures advocating for AI in selecting students for admission to university courses. Imagine an AI bot trawling through your social network data, your psychometric test results, your resume, and the transcript of the interview to determine whether or not you are a good fit.
This isn't science fiction. Amazon already tried to use AI in their recruitment process. As Alexa herself will tell you, Amazon is a world leader in the implementation of AI, but they got this one wrong. Their results tended to discriminate against women, and the program was scrapped.
However, it's certainly not the last we will hear of this. Deep neural networks need a huge amount of data to draw on if their decisions are to be accurate, and so it is only a matter of time, and not necessarily know-how. IBM's Watson is already the front runner for the application of AI to the student selection process, and with their resources behind it, we are looking at a real shift in this direction in the not too distant future.
Though it may seem creepy to be "artificially" selected as a candidate for an undergraduate degree, there is a flipside in that AI has no moral agency, no consciousness, and therefore, no bias. Humans, on the other hand, carry deep-seated unconscious bias that has plagued the world of admissions for so long that it has become almost normalized. Perhaps AI can help us to be truly objective?
True AI-powered chatbots are probably the best-known frontline use of AI in education marketing and recruitment. In 2018, a special bot built for the University of Murcia was able to respond to students' questions with an extremely high rate of accuracy, as well as dealing with over 800 enquiries in one single day outside of office hours; a huge advantage in international admissions, dealing with numerous time zones simultaneously.
Most AI bots now will use learnings from pilots such as this to determine where the enquiry needs to be passed to a human to deal with. Companies like Conversica are working hard to make sure that learning is continuous on the nuances of human communication and that politeness is always the approach. If you manage to make an AI chatbot angry, then call your therapist immediately.
The study author from the Murcia chatbot pilot observed that no humans lost their jobs as a result of the bot's handling of huge amounts of enquiries, and that admissions team members were instead freed up to spend more time building relationships, solving more complex issues and following up on things they may have otherwise missed in the hectic workflows of a busy university. That being said, there are naturally a few legitimate concerns around the use of AI in education marketing and recruitment.
Both Gen Z and the forthcoming Gen Alpha are reputed to be less concerned about privacy than the millennials before them. They expect personalization and customization, and only a data-driven service can offer that.
But that is not the whole story. AI only works well with access to huge amounts of data, and because AI evolves in the way it analyzes and interprets data, it will also, therefore, create new personal data. The user ticks the box or signs the consent form to use the data that they initially provide the institution, but if new data is created through AI that can be used to identify or target them, this is a clear breach of data protection laws in many countries.
In such a situation, the institution would have to contact the student to request permission for the new layer of data, and who really knows where this new generation will draw the line. In a more general population, nearly two-thirds of respondents to a Genpact survey in 2018, said they were uncomfortable with AI being used to make decisions using their data without their knowledge. This is clearly a situation that developers will find challenging.
What we do know, is that with both Gen Z and Alpha, when they are interacting with our institutions, their perception of purpose is critical. If they believe in what you do, the causes you support, the values you uphold and embed, then they are more likely to be flexible with the information you hold on them and how it is used.
For years now, AI in Education Marketing and Recruitment has been ever more present. We have seen first hand just how much more efficient an institution becomes with the right kind of automation and the training to support it.
AI is a logical extension to this, giving a level of personalized service that can only be good for students making the important decision about where and what to study, and taking the strain off an increasingly stressful environment for staff.
Where the technology is used to augment the experience for everyone and not to replace humans with bots, we're all in. Where data is used with consent to provide an enriched experience for those interacting with our institutions, we're in. Where AI in Education Marketing and recruitment goes beyond this is uncertain, but next week we'll be looking more broadly at its application to teaching and learning, so check back on our blog for more on this fascinating field.
