Even before the launch of ChatGPT in November 2022, many employers placed a high value on skills and experience in artificial intelligence (AI). In fact, 80% of employers in the education sector told the QS Global Employer Survey in 2022 that they felt that AI was important to their organisation.
This article explores the growth in the use of machine learning – across a range of sectors and industries including how universities are using the technology to convert student recruitment enquiries and offers more effectively.
What is machine learning?
Machine learning (ML) – a subfield of artificial intelligence (AI) – is a technology that allows machines, such as robots or computers, to learn, adapt, perceive and make inferences using statistical algorithms, geometry and probability theory.
The general use of machine learning is used to increase automation and efficiency, enabling data-driven decision-making and greater personalisation of products and services.
As McKinsey’s Global Survey in 2022 reveals, the adoption of AI technologies more generally is continuing to rise – with more companies than ever before leveraging AI models with ML functionality, such as ChatGPT.
According to McKinsey, 56% of all organisations surveyed in 2021 reported AI adoption in at least one function – up from 50% in 2020 – with this increase seen most in emerging economies such as China, the Middle East and North Africa. In fact, adoption of AI has more than doubled since 2017.
The increased interest in and usage of AI models that rely on ML in recent years can be attributed to a range of factors. This includes the availability of large amounts of data and powerful computer resources, such as Graphics Processing Units (GPU), which have led to very powerful AI functionality that helps humans navigate everyday life more efficiently.
How is machine learning used today?
While the use of ML increases, so does its demand in a wide range of industries:
Healthcare: ML is being used in the medical industry to help ease its pressure on increasingly in-demand helplines and administrative services by leveraging algorithms to better manage patient records or schedule appointments. ML is also utilised to find signs that indicate a specific disease through medical imaging (such as X-rays or MRI scans).
According to Dr Roman Bauer, lecturer in Computer Science at the Centre for Mathematical and Computational Biology. University of Surrey: “AI will soon revolutionise medicine by rendering the diagnosis and monitoring of patients’ diseases more efficient, reliable and fast, ultimately helping decrease the increasing burden on healthcare systems all over the world.”
Finance: Some of the most widely adopted applications of machine learning in finance include fraud detection, risk management, process automation (such as email and browser automation), data analytics (to predict market trends), customer support and algorithmic trading (such as for reading current market prices or to reduce trading expenses).
Marketing: ML is used in marketing to predict the behaviour of customers by finding patterns in their online journeys. It can automatically identify objects, people and even emotions in images and videos, allowing marketers to deliver more relevant tags, captions, adverts and recommendations.
“In the case of marketing, targeted publicity is now very common in many social networks and we expect to see more of this type of application in the near future,” explains Dr Victor Sanchez, Associate Professor of Computer Science at University of Warwick. He cautions: “However, it’s important to always remember that these technologies should be used in an ethical and unbiased manner.”
Education: ML can be used to create personalised learning pathways for individual students – increasing the quality and efficiency of teaching methods in accordance with each individual’s learning style. Using an AI and learning platform Century Tech, for example, students begin by completing diagnostic assessments, which reveal learning gaps and areas for improvement. With AI-powered features, Century recommends topics that students need the most help with, while reintroducing content at regular intervals to prevent students from forgetting what they’ve already learned.
How can machine learning help institutions convert student enquiries, applications and offers more effectively?
To better understand how ML works in the context of student recruitment, let’s look at an ordinary analogy. If an individual regularly searches for innovative cooking recipes on a social media platform such as Pinterest, that platform will provide more recommendations of recipes to try beyond initial results. These recommendations are essentially predictions that have been informed by previous user behaviour from across the platform and form the core of a user’s ‘Picked for you’ category.
In a similar way, ML models can utilise historical and current institutional data, on the behaviour and demographic of student applicants and prospects, to identify the most significant factors that influence the journey along the student pipeline. This information helps institutions understand where and how to put their efforts when it comes to converting enquiries to applications and offers to enrolments.
Although machine learning cannot perfectly predict every outcome, it certainly helps institutions make more informed and intelligent choices in their student recruitment efforts – something QS has witnessed firsthand through the deployment of its bespoke machine learning functionality.
Speaking on the information used by the QS model, Ha Ho Hai, QS Director of Business Intelligence, explains: “We use a combination of demographic (age, country, course), communication preferences (engagement with email or phone calls) and behavioural (levels of satisfaction and engagement with communications). Models may look at up to 200 data points about individual enquirers or students and they are refined throughout the recruitment cycle.”
What are some of the benefits of using machine learning for student recruitment?
In our report, ‘Machine learning: How UK universities are improving quality and conversion rates’, we investigated how institutions were navigating the significant increase in student enquiries and applications seen in the last few years: our UK partner institutions experiencing an 82% rise in new enquiries between 2021 and 2020, with the number of new applications growing by 179% in the same period.
In higher education, just as in most sectors and industries, AI offers opportunities to deliver efficiencies at scale, particularly when services are in high demand. Admissions and recruitment teams can focus their efforts on those enquiries and offer holders that are more likely to convert, using ML models to simplify, understand and interpret large data sets more efficiently than a human could.
QS partnered with several UK universities to improve student recruitment conversion rates for September 2021. By utilising QS machine learning models, the University of Stirling almost tripled their conversion rate.
In addition to the volume of applications and enquiries, prospective students in 2022 are expecting faster responses from institutions. In another QS report, ‘How universities in Australia and New Zealand are using artificial intelligence to qualify and convert’, we revealed how 74% of prospective international students interested in studying in New Zealand or Australia expected a response within a week of applying to a university. A staggering 92% of respondents expected a complete and personal response within a week – 32% within just 24 hours. Often, admissions teams are unable to meet these challenging timeframes due to their already overstretched staff.
QS Business Development Director Corey Thiedeman explains why recruitment teams must rely on advancing technology to meet growing expectations targets in 2023.
“In Australia, recruitment teams face a challenging year ahead with huge increases in enquiries and applications. New Zealand may also face a growth in volume during 2023 as universities reap the rewards of working hard to rebuild their momentum in key international markets. With restructures and redundancies resulting from the decrease in international student tuition fees during the pandemic, many universities are operating with far fewer staff and on significantly reduced budgets. It’s a perfect storm which is risking burnout from dedicated colleagues working in international recruitment and potential reputational damage in the market from slow turn-around times in response to enquiries and applications.”
Far from replacing recruitment and admissions teams, ML enables those teams to refocus their time doing work that cannot be automated – for example, counselling prospective students and delivering conversion campaigns.
So, will machine learning ever replace human intelligence?
When humans learn by experience, they rely on a range of abilities including muscle memory, emotional memory and intuition. Take a human trying to walk across a tightrope – they may fall at first but, with practice, they will eventually reach the end of the wire. However, for ML learning technologies to perform certain tasks, they first require programming and instructions.
Dr Hyung Jin Chang, Associate Professor of Computer Science at the University of Birmingham, elaborates “Fundamentally, current ML can perform well only on specific, well-defined tasks such as image classification, image creation, translation, etc. While a lot of attention is being given to whether ML will achieve human-like comprehensive intelligence in the near future, we must remember that solving specific tasks better than humans does not mean that it is ‘intelligent’.
“Since machine learning relies on refined data converted into a form that computers can understand, I don’t think its intelligence will match that of humans just yet. First, the amount and types of data that humans accept and process are particularly large and varied – not to mention the amount of continuous learning through interaction with the world that humans experience.
“Secondly, it cannot be ignored that the human brain analyses and accumulates experiences and sensory data particularly effectively and efficiently. Its ability to learn is superior compared to any state-of-the-art computers.”