Alexa, tell Bolton College I’m being bullied.”

It is 3am. Grace* hasn’t slept a wink; in just three hours’ time, she has to get up to go to college. But she can’t stop the thoughts about what the day will bring racing through her head.

Speaking the above words to her bedside digital personal assistant, Ada, sets off an instant chain of events - hours before the first member of staff even arrives at the college.

Straightaway, Grace’s smart speaker tells her reassuring information about the support that her college offers through its anti-bullying procedures.

An email is automatically sent to notify her college’s safeguarding team, too, so action can be taken to start resolving Grace’s situation first thing in the morning, before she has even set foot on campus.

In recent years, the use of AI in education been spreading. But experts believe that further education could have more to gain from the tech revolution than most.

From a simple chatbot developed to answer students’ questions to an innovative programme that uses data about individual students’ strengths and weaknesses to enable teachers to support them through GCSE resits, AI is already beginning to transform the FE sector.

Bolton College’s Ada platform is named after Ada Lovelace, the 19th-century mathematician often referred to as the first computer programmer. She worked on a proposed mechanical computer and, as the first person to recognise the true potential of such a machine beyond being purely a device for calculations, developed a prototype algorithm. Her work helped to inspire other pioneers such as Alan Turing and his team of codebreakers at Bletchley Park during the Second World War.

Artificial intelligence in colleges

Ada can be accessed through a website, a mobile app or a voice-activated assistant. It can answer more than 12,000 variations of questions and problems, including everything from “I need a new bus pass” to “How do I do a fashion perm?”

“It’s like walking up to a reception counter or walking up to student services and posing a verbal question to the desk,” explains Aftab Hussain, Bolton College’s strategic information learning technology (ILT) lead. “But in this case, rather than the receptionist or the careers adviser being behind the counter, you have the whole college ready to answer a question, because behind Ada you have got the 600 staff at the college.”

The platform incorporates “nudges” - which means it proactively sends information out to students. This can range from a simple “Welcome back to college” to more serious cases in which, if a student’s attendance slips, they will be sent a message encouraging them and also offering them support.

“Imagine you’re a fragile learner and you haven’t done so well at school, and you are about to start your studies at a large college and you have no idea where to go on Monday,” Hussain says. “If you’re a fragile learner, you just might not turn up on a Monday morning if you didn’t have an answer to the questions ‘Where do I need to be?’ or ‘Who is my teacher?’ But because Ada can give them the answer - because it is connected to their dataset - then hopefully that student will say, ‘Yeah, I know where I need to be on Monday,’ and then off they go to their studies.”

During the summer term, students will be sent reminders about their exams. Not only will the programme let learners know where their exams will be, but it also knows what their grades are like and it can suggest advice accordingly.

“It is more than just a notification. It will offer advice and guidance about what you need to do to maintain and increase your [grade] averages,” Hussain explains.

Since the platform launched two years ago, student retention at Bolton College has gone up by 4 per cent. This has had a particular impact in the first half of the autumn term - which is critical for colleges in terms of keeping per-student funding.

“We can’t say Ada is solely responsible for that,” says Dean Baggaley, systems development leader at the college. “There are many contributing factors and it’s very difficult to quantify, but what we are doing is putting intelligence information in simple terms, which saves people time.”

Ada also helps the college’s staff and teachers in their daily activities. For example, staff could use the platform to ask, “Who is on work experience this week?” This would reveal which students will be out of lessons, where they are going and what the contact details are for staff there. They can also find the answers to simple HR questions, such as: “Am I in the pension scheme?”

Staff can even programme information from their classes into the app - so Ada can answer questions such as “How do you simplify ratios?” or “How do you make cement?”

Viv Fernside, training and development manager at Bolton College, says that, despite some initial scepticism, there has been a “huge input” from staff, making Ada relevant across curriculum and business areas.

“For staff, it gets rid of a lot of white noise - the constant same questions that we get asked time and time again. You just have to point students to Ada,” she says. “It frees up teachers to do more teaching and the things they need to do rather than [answer] the constant same enquiries.”

But the potential of AI is greater than simply creating efficiencies. South and City College Birmingham is using the Century platform to develop personal learning plans for its GCSE resit students.

Century is a social enterprise that has developed a learning platform based on artificial intelligence. This use of AI goes way beyond conventional edtech, explains chief executive Priya Lakhani.

“Edtech has existed for a long time, where you have your virtual learning environments, for example. AI is a machine that makes smarter decisions based on the information that it is collecting,” she says.

“These are beastly sorts of systems; they are huge. They require a lot of development work and they are not built by web developers; they are built by data scientists, data engineers and learning scientists in education.

“The machines collect data on user behaviour and on assessments and then the machine looks for patterns and correlations in how people are behaving, and can change the next step and change what people see based on those patterns and correlations. Although humans built these systems, actually they become smarter and smarter and grow themselves.

“It can also be a machine that will predictively forecast how you are going to behave. In a college, it could look at how all the students are behaving with maths and English resits. This platform could come up with a dashboard with all the at-risk students within a few weeks of them using the system so the college could make interventions.

“Using all of the data from previous cohorts about how they engage with maths and English, and based on all the interactions with the current cohort, the machine can forecast that these ones are likely to drop out.

“The machine is starting to be intelligent about patterns and correlations in behaviours so it means the college can intervene.”

The use of the Century platform has led to changes in the delivery of GCSE resits at the college. Teachers now focus more on interventions when students are struggling, rather than delivering a lesson from the front of a classroom.

The platform uses AI to learn a student’s strengths and weaknesses, as well as their learning pace and preferences, to develop a path towards mastery of a given subject. A live dashboard shows the teacher exactly when a student is finding a topic hard, so they can then engage with them and guide them through.

Alice Little, senior cognitive neuroscientist at Century, explains: “I think what’s really interesting in the FE context, compared to in the school context, is that a lot of these students are coming into college and - for whatever reason - the way they were being taught or experiencing English and maths was not working for them. So they can feel really disheartened going into that resit classroom. They can feel like that learning is completely closed off from them.

“In school, they might not understand a topic from [back in] key stage 2 or 3, so those emotions about ‘feeling stupid’ can be reinforced.

“With this platform, we have content that is deliberately designed for older students, and it doesn’t feel babyish or patronising.”

The platform will remind the learner of concepts through algorithms developed by Century, which used neuroscience research on memory function to look at the optimum time in the learning process for a concept to be reinforced to stay in a learner’s long-term memory.

Lakhani adds: “The whole point of Century is it will figure out why you are struggling, so it won’t give you Pythagoras’ theorem if it knows you don’t understand roots and powers, for example.

“[When students are ready] they realise: ‘Oh, I am doing Pythagoras’ theorem’. They just needed those stepping stones to get there. What the machine does is it means people don’t have to put their hand up in the classroom to say, ‘I don’t understand.’ Not all learners will do that, and some of them will actively shy away from it because they feel embarrassed or humiliated that they don’t understand when their peers might be progressing.

“It’s your own personal experience [through the platform] that will provide you with a path to mastery, but it will do it in a way where it will learn exactly what you do and don’t know. The nicest feedback I had about the platform was from a student who said it didn’t make her feel stupid.”

Fears that AI will replace teachers in the classroom are well-documented. Last month, John Hattie, director of the Melbourne Education Research Institute, warned teachers at a conference in Edinburgh not to underestimate the ability of robots to take over at least some aspects of their job, saying that he recently observed a robot teaching a class, after which the pupils said they preferred it to their teachers.

But Lakhani is quick to affirm the important role of the teacher. “We have a fundamental belief at Century that the most important person in the classroom is the learner, but the most powerful person in the classroom is the teacher,” she says.

“We spent lots of time in classrooms and thought, ‘Why is the teacher spending 60 per cent of their time on admin?’ ‘Why do we also have colleges where there is a psychology graduate teaching maths?’ We are putting an awful lot of pressure on these teachers to succeed, but we’re not giving them the tools to succeed.”

AI teaching platforms can help to reduce teachers’ admin so that their face time with learners is about targeted and individualised interventions - in accordance with the flipped learning model.

Lakhani adds: “This is a really powerful way of teaching that has been proven and tried and tested. The most powerful thing teachers can do is make interventions and guide their learners through the curriculum - particularly if it’s not a teacher who is qualified in that subject. So when a student struggles or stumbles, then they can make the intervention, rather than spending one hour of a two-hour lesson dictating a one-size-fits-all delivery of education, where some learners will not be understanding and some will be needing stretch.

“The teacher is no longer passive - they are stood there dictating or looking through books, trying to work out what people know and don’t know. The teacher has an active live dashboard that shows them instantly which students need support and where they need support and whether they are engaging. So, what you have is a very efficient classroom.”

At Bolton College, the idea is not to replace real-life workers but to make people’s jobs easier. “We’re never trying to remove people’s jobs or anything like that - we’re always trying to supplement people’s work and make life easier for them,” Baggaley says.

The team is also developing more cognitive services, including an automatic marking system and the automated production of reports.

Hussain explains: “We have got 2,000 students going out on work experience a year and only a handful of staff to evaluate those placements. It’s a problem that can be automated. The machine we’ve created can read, understand and classify the information.

“Ada could also inform a student what they need to do to push to get a higher grade and act as a nudge.”

But this could be just the beginning. Hussain’s ultimate vision is of a world where a digital assistant follows a learner throughout their educational life - irrespective of what institution they are studying at.

“The idea is that every student, every teacher has access to their own digital assistant. From preschool right up to graduation at university, or through work - you’ve still got your digital assistant supporting you. I love that whole idea that you’ve got this companion who is supporting your learning throughout your whole lifetime.” He pauses. “Imagine how it would change education.”

*Grace is not a real student

George Ryan is a Tes FE reporter

This article originally appeared in the 12 April 2019 issue under the headline “Rise of the machines”