AI technologies can make instruction more personalised, therefore increasing the effectiveness of education

Education around the world is changing at an incredibly fast pace. Software and devices are as commonplace in classrooms and lecture theatres as whiteboards and overhead projectors once were. The new generation of student is born digital and they are fast adopters of new technologies. The revolution has begun.

Education institutions globally are grappling with three big challenges: Providing quality education, often at scale; ensuring education is accessible to all, including in emerging markets, rural communities and to children with special needs; and reducing the cost of delivery to provide affordable education.

Teachers are burdened with administrative tasks that take time away from teaching: Planning class materials for large classes of mixed ability students; marking and grading assessments and homework; and fact and source checking for submitted assignments. School administrators and admissions staff, meanwhile, are struggling with selecting the best students from a large volume of applications and effectively communicating with students, staff and parents.

All of this is resulting in an education system that is under-resourced and inefficient. And where all too often, students can get forgotten and left behind.

There are a multitude of edtech solutions appearing daily, but undoubtedly one of the most exciting technologies that has the potential to have the largest impact on education is Artificial Intelligence (AI). With the increasing sophistication of AI techniques such as natural language processing, voice and speech recognition and machine learning, education instruction and administration can be transformed.

A recent Pearson report summarises this nicely: “The future offers the potential of even greater tools and supports. Imagine lifelong learning companions powered by AI that can accompany and support individual learners throughout their studies – in and beyond school – or new forms of assessment that measure learning while it is taking place, shaping the learning experience in real time.”

So what problems can AI solve? Let’s take a look.

Standardised curriculum does not cater to individual needs

Every student is different. Some learn faster than others, some are able to grasp more difficult concepts than others, but we expect them to conform to one basic curriculum. Personalised learning can change that. Students work their way through tasks at their own pace, and the platform learns with them, adapting to their needs and ensuring comprehension.

It encourages them to work independently and manage their own learning, skills that are essential for the modern workplace. The platform can also flag issues to teachers, so that they can reinforce learning directly with the student. The ultimate aim is to give students more control over what and how they learn. Research by The Gates Foundation has revealed that over a two-year period, students using personalised learning software had more success in their studies than their peers who do not use the software.

Also Read: A look into China’s future: Unravelling China’s edtech landscape

At the Aspen Valley Prep Academy in California, a personalised learning system helped move below-grade level students forward, by delivering tailored content according to their skills and proficiencies. Another example is the Urban Assembly Maker Academy in New York City, which now has 70 per cent on-time work submission, following the introduction of a personalised learning initiative.

Oregon State University used an adaptive learning process, where 8 courses in disciplines such as algebra and chemistry, with large enrolment and high attrition rates, adopted a blended learning strategy where students completed activities before their classes. These activities were then analysed and instructors then adapted their teaching where needed. McGraw Hill Education found that 85 per cent of students who were interviewed felt that adaptive learning technology was a ’moderate or major improvement’ towards better grades.

Grading & assessment is time-consuming, with an over-reliance on multiple choice

On average, teachers spend over 11 hours on marking every week; that’s more than an entire working week each month, on top of their teaching duties. In an effort to reduce this burden, many institutions now use multiple choice assessments, which are less-effective at testing overall learning. AI-powered assessment tools, such as that under development by Sergey Karayev and his team at Gradescope can remove some of this burden, with automatic essay scoring technology, reducing the reliance on multiple choice assessment and ensuring better student comprehension.

Limited one-on-one tutor time available for university students

Most university teaching is delivered in large-scale lectures to hundreds of students at the same time, with little time dedicated to even small group teaching, let alone individual teaching. Personal virtual tutors, such as that under development by NTU Singapore & IBM Watson for its medical students, could change that, delivering personalised, scalable guidance to students.

Large class sizes in K-12 schools means children’s questions often go unanswered

Teaching to classes of sometimes over 30 students, means that teachers have little time to constantly reinforce learning and answer individual questions from every student. Using virtual assistants in the classroom, such as Amazon’s Alexa, as Hillbrook School in California have trialled, could change all that, offering students the opportunity to check their own answers, freeing the teacher from bottleneck questions such as “How do you spell…”. It also empowers teachers with analytics on the questions that have been asked, offering them additional opportunities to identify weaknesses in comprehension and modify their teaching accordingly.

Personalised communication is almost impossible due to scale

Giving students, parents and teachers access to the information they need quickly and easily is a huge challenge. According to a 2014 Gallup survey, millennials cite text messaging as their primary method of communication, so digital messaging has enormous potential. This could also result in significant cost savings. Taking this further Georgia State University used a chatbot to communicate with its students. In just 4 months, 63 per cent were using the platform on a regular basis, with over 200,000 messages sent. They found this would have required 10 full-time staff, saving them US$200,000.

Student enrolment

Georgia State University had a problem where students who were successful in their college admissions failed to enrol as planned after their summer breaks. Whilst they knew that personalised communication with the students would help reduce this ‘summer melt’, they did not have the resources to do it at scale. With the help of AdmitHub, they created Pounce – a personalized virtual assistant to send timely reminders and relevant information about enrollment tasks, collecting key survey data, and instantly answering students’ many questions around the clock. Student engagement was high with 90 per cent opting in and the final result saw enrollment increase by 3.9 per cent.

Student retention

The University of Oklahoma, like many universities, was finding student dropout rates challenging, losing around 14 per cent of students between the first and second year enrolment in 2015. To reduce dropout rates, the university worked with IBM to develop a system which helped identify students more likely to drop out before they started the first year. Through IBM Watson’s sentiment analysis technology (such as personality insights and natural language classifiers), they were able to offer a more personalised experience and earlier intervention, increasing their retention rate to 92 per cent in just 2 years.

Effective combating of plagiarism and ensuring authorship

The Josephson Institute Center for Youth Ethics found that 33.3 per cent of the 43,000 public and private high school students surveyed admitted to plagiarism, whilst research by Rutgers University revealed 7 per cent of undergraduate students and 4 per cent graduate students plagiarise word-for-word without citation. AI-powered software such as Emma Identity will help counteract this, identifying not just direct plagiarism but also authorship based on writing style, vocabulary and more.

We have summarised these applications in the below diagram:

Risks & Limitations

There are several risks & limitations with the implementation of AI technology:

Poor teacher adoption: Some teachers will be resistant to this and will need to be managed through the change accordingly
“Fear Factor”: As with many industries teachers may feel that AI technology could ultimately replace them and will therefore be reticent to push forward initiatives using the technology
Adoption at scale: AI integration requires support from multiple stakeholders – parents, students, teachers, administrators and policymakers. This can be difficult to achieve quickly
Subject limitations: AI won’t be relevant to all subjects. For example, personalised learning and automated grading is unlikely to work in practical subjects with a strong subjective element to assessment, such as drama, art, food technology and physical education.

Investment: With schools under more financial pressure than ever before, deployment of AI technology at scale is expensive and not all schools will have easy access to such funds.

Privacy, data & cybersecurity: This is likely the single biggest risk of AI deployment in education as so much personal data is needed for success. In the education environment, this challenge is compounded further by the fact that you will, in many cases, be dealing with minors’ personal data and information, which is governed by tighter legislation. A robust cybersecurity and data privacy strategy will be integral to success.

Also Read: Edtech startup Cudy turns your smartphone/laptop into a classroom; offers live tuition

Virtual assistants: The use of virtual assistants such as Siri and Alexa may offer outstanding learning opportunities, but the laws surrounding what data these assistants are collecting and how it is used are not fully developed. Before this technology can be meaningfully implemented in schools (outside of students’ own personal devices), this will need to be addressed by regulators and governments.

Recommendations

As illustrated above, the innovative applications of AI in education are many, and they are applicable both from a student learning perspective (e.g. personalised learning) and from an institutional perspective (e.g. grading & assessment).

Given the vast potential implications of AI adoption in education, we recommend focusing initial investment and effort on two of the identified applications – personalised learning and grading & assessment. We believe that these two applications bring significant benefits: improved learning for students and reducing the administrative burden on teachers, freeing then up to spend more time on real teaching.

To accelerate the pace of change, we recommend that educational institutions partner and collaborate with technology vendors to develop solutions that fit their individual circumstances and best serve their students and staff.

Of course, educational institutions are also free to not adopt AI technology in any way, however, we feel that this would be short-sighted. The University environment is hugely competitive, with large fees at stake. Those who choose not to integrate AI technology will likely find themselves lagging behind the competition, resulting in lower student enrolment and revenue. At a K-12 level, not embracing AI solutions will mean education is restricted to a one-to-many approach, with large administrative burdens on staff and little opportunity to deliver quality teaching to students.

Interested to learn more about how AI is transforming education? Join us at EduTECH Asia 2018 to find out more.

Join us at EduTECH Asia 2018

At EduTECH Asia, we pride ourselves on bringing together thousands of educators from across Asia to evaluate and plan for the future of learning. In 2018we’re be going bigger and better than ever before. Spanning across 3 exciting days from 8-10 October at Suntec Singapore, the event will feature 8 theatres of content spanning topics in K-12 Education, Tertiary, Technology, Special Needs, Vocational, The Business of Education, Early Childhood and, new for 2018, EduBUILD. Plus, there will be workshops, EduSHARE roundtables, EduSLAM Sessions, and more. Supported by IMDA, event will also feature an exhibition showcasing the latest edtech innovations and technologies. Find out more!

—
Copyright: rawpixel / 123RF Stock Photo

Editor’s note: e27 publishes relevant guest contributions from the community. Share your honest opinions and expert knowledge by submitting your content here.

The post How artificial intelligence is disrupting education appeared first on e27.