At our annual edtech showcase Digifest, 12-13 March 2019, delegates can experience the latest technologies and learn from edtech experts. Rose Luckin, professor of learner centred design at University College of London, shared some of her hopes and fears for artificial intelligence (AI) at last year’s event.

When I look at what is possible with AI, my optimistic side – which usually takes over – sees the huge possibilities it brings for people all over the world, both for those who are privileged to have experienced a good education and those who have not.Universities have put much effort into widening participation and AI technology has been discussed as a means of helping those hardest to reach, including adult returners and others who don’t live on campus.  What AI can bring to the classroomThere are many benefits that AI can bring to teaching and learning. It is great at helping us to analyse data and allow us to better understand students. It’s consistent and doesn’t make the mistakes a human being could. If designed in a way that is informed of our knowledge of how people learn, it can adapt well to an individual learner.[#pullquote#]Well-designed AI that uses machine-learning improves the more it supports learners, so it is constantly getting better at individualising support effectively. [#endpullquote#]Well-designed AI that uses machine-learning improves the more it supports learners, so it is constantly getting better at individualising support effectively.But AI has its limitations. Humans are way more sophisticated in their intelligence. For example, being able to accurately know what we don’t know, as well as what we do know is important and it is part of our metacognition.AI can mimic emotional intelligence, but it is not emotionally intelligent – a teacher teaching one-to-one is still the gold standard.AI teaching assistantsSo, AI can’t replace teachers. Although I can see a role for an AI teaching assistant.I can imagine that for budget-constrained decision-makers, an AI tutor that is never off sick, never goes on strike, is always accurate and consistent and can deliver individualised content in the core curricular subjects sounds very attractive. The initial outlay cost is high, but the ongoing costs are low.The combination of being able to collect masses of data as we interact in the world, and smart AI algorithms that can process this data to reveal the nuances of our learning progress across a whole range of skills, abilities and knowledge means that we really can start to shine a light on people’s talents, beyond those we traditionally value.This opens the door to people who are currently not able to demonstrate adequately how they are progressing.Supporting students with AII was a school “refuser” at 14. I didn’t go to university until I was 32, so I have a little bit of understanding of what it’s like to be that kind of “non-traditional” person that doesn’t get on a straight and narrow road at an early age.If we move to a system whereby everybody was provided with their own AI assistant that provided AI-driven, continuous, formative assessment and support, then someone who hasn’t received teaching that would allow them to pass an exam could, with the right support, be able to demonstrate their skills and abilities.For example, if that person had overcome a lot of challenges and had the evidence and data to demonstrate that to an employer, it could give them the edge over someone who has had an easier path.By demonstrating a richer set of knowledge, understanding, skills and abilities than we currently can, such data would be valuable for learners from lots of different backgrounds.Could this lead to further division?But I worry about the extremely dystopian possibility – which is a technical possibility – that AI could exacerbate the divisions that already exist in education.I can foresee a situation where we have classrooms and lecture halls with lots of young people interacting with these AI systems, while only the more privileged keep the human interaction.[#pullquote#]if used badly, AI could actually make inequalities worse.[#endpullquote#]There is also a risk that, if used badly, AI could actually make inequalities worse. Just look at the way in which we have allowed many people to own sophisticated technology, whether a television or a mobile smartphone, without necessarily enabling them to use it to better their chances in life.Communicating AI’s benefitsWe must find ways of communicating effectively to everybody what the technology can do and how best it can be used safely to better their chances.[#pullquote#]The education system is short of money and short of human teachers, so I think it is something that we should be particularly vigilant about. [#endpullquote#]The education system is short of money and short of human teachers, so I think it is something that we should be particularly vigilant about.It’s up to educators to get the conversation going with the tech companies and teachers about AI in education. We do it in a small way here at UCL with our EDUCATE project, which is all about getting edtech companies to talk to educators, researchers and students. But we need to make that happen elsewhere.We need AI developers and teachers to work together to co-deign the AI systems for education. EDUCATE has developed a model for this type of inter stakeholder collaboration.We then need to focus our education system on helping people to become knowledgeable and skilled at the things that are not possible for AI to do. A mass AI education programme for anyone, whether they are in university, college, school, work or out of work, will enable people to understand how to protect themselves against competition from AI and to be able to use this technology wisely.The description of AI and machine learning as the fourth industrial revolution is not wrong. It is a revolution and we need to prepare people for it.This blog is based on a question and answer session following Rose Luckin’s presentation at Digifest 2018. Read about her vision for how educators can draw on AI in her book, Machine Learning and Human Intelligence.Booking is now open for this year’s Digifest, 12-13 March 2019.