One of the most popular “How” arguments in artificial intelligence for education relates to personalised learning. Essentially the argument goes…
“AI will allow us to personalise learning for individual students and get away from the one-size-fits-all approach”
But what will this look like? What’s already happening and is it working?
Microsoft founder Bill Gates and Facebook CEO Mark Zuckerberg are both strong proponents of the idea that AI can personalise learning. Both point to a number of existing and emerging technologies that they say promise to improve outcomes for students by offering them experiences and pathways that cater to their specific needs. Both are investing through their respective foundations in efforts to see personalised learning solutions developed for schools.
Learning and class management systems
Proponents of AI for personalised learning often point to online systems like Blackboard, Google Classroom and Edmodo that allow teachers to collect, curate, combine and communicate content to students in ways that are customised for individuals and groups. This allows teachers to break the traditional uniform delivery mode of traditional teaching and help students work on the specific concepts they need and/or are ready to engage.
It is debatable whether this really involves or requires artificial intelligence though – this kind of functionality has existed for at least 20 years and the personalisation is really coming from the teacher unless one of the other concepts (e.g. below) is also in play.
This is where systems use algorithms to automate the collation of learning experiences for students based on their demonstrated learning needs. In the Australian assessment system Improve, for example, students can undertake a teacher-created test and then receive learning activities the system determines will help them improve in areas of underperformance.
It has been argued that software could be developed that also takes into account students’ individual learning styles – notwithstanding the doubt over their existence.
One of the propositions of AI in education is that software could assist teachers and students with the often laborious and complex task that involves mapping learning experiences to educational standards such as Common Core (USA) or the Australian Curriculum. In the Learning Place and Scootle, algorithms already help teachers match curriculum outcomes with relevant learning resources (and back again) and more intelligent systems could conceivably analyse more complex and authentic learning experiences (e.g. project based learning) to map to standards.
Personalised AI Tutors
A seemingly far-fetched idea is that an AI enabled tutor could be used to offer advice and feedback to learners, not just for curriculum areas that involve right-wrong answers but in subjects that normally involve higher-level conceptual skills and understanding such as writing. AI is already used in journalism to create news reports that have been shown to be hard to distinguished from human-generated ones, so is it too hard to imagine software that can analyse a student’s writing and offer feedback? This argument also points to the value for learners who might be reluctant to seek feedback from peers or a human teacher although some thought about what a teacher’s role would be, and how the teacher-student relationship would/should evolve, is probably needed.
Reporting, accountability and home-school engagement
If curriculum planning and delivery can be personalised through AI, then it is arguable that assessment and reporting can also be individualised with much of the process automated to reduce rather than increase workload on teachers. However, it will probably require a mindset change away from standards as a tool for ranking teachers and schools, to the use of standards and AI as a way of creating more effective learning partnerships between schools, their students and families.
(Also published on the Learning Place 18 April 2017)