The recent flurry of discussions about AI, cheating and the purpose of education have got me reflecting on Michael Trucano thought-provoking post from 2023 about how artificial intelligence might change the evolution of digital divides in education. Mike’s central proposal is this:
- The first digital divide: The rich have technology, while the poor do not.
- The second digital divide: The rich have technology and the skills to use it effectively, while the poor have technology but lack skills to use it effectively.
- The third digital divide: The rich have access to both technology and people to help them use it, while the poor have access to technology only.
I fully agree with his take on the third divide – but is there more to consider two years later? Below are some personal ‘thinking out loud’ observations. Please share your reactions and any research to strengthen, refute or add to the points.
Is the rich/poor differentiator still valid?
Even in rich countries the people who should be guiding the use of AI are not stepping up. The sobering article from the New Yorker magazine Everyone Is Cheating Their Way Through College describes how at top US universities AI is used by students to do assignments, and by faculty to grade them.
This points to a future where AI does the work and grades itself, and people just organize the process. In the US, 86% of higher ed students are using AI in their studies (we may be the last generation who write original, unassisted text!) – yet 58% don’t feel they have sufficient AI literacy.
In the last year, the number of higher-ed instructors that use Gen AI tools frequently has gone from 18% to nearly double that.
What about those in developing countries?
The New Yorker article asks a key question: “Who could resist a tool that makes every assignment easier with seemingly no consequences?” (It goes on to outline the very serious consequences.) What if we ask this question at the school level and in the global South – even in contexts of ‘learning poverty’, where 70% of 10-year-olds are unable to read and understand a simple text?
If assignments and grading are outsourced to AI in well-resourced settings, is the tool more irresistible here?
In many developing countries, teachers face significant challenges: very large classes, limited pedagogical skills, insufficient subject matter expertise, and low motivation. These issues often stem from inadequate support and poor school infrastructure.
AI uptake amongst children is already high, e.g. reports this month from Argentina show 58% of children use ChatGPT, including 37% of 9-11 year olds, mostly for homework support. UNICEF Innocenti’s research (forthcoming) with 9,000 teens around the world found ‘helping with homework’ to be the most common reason for AI usage.
AI for cheating just gets easier
Over the years, Mike and I have often agreed that the way edtech scales is when you leverage the tech already in people’s hands. In the last 15 years that has meant through mobile. AI on mobile has fast-tracked this: Using WhatsApp’s AI as a study buddy, within a minute I got it to do a short writing assignment (as the student), and grade it (as the teacher).
It gave itself an A! (Caveat: this was done in English. I asked for Zulu, but no can do for now).
The app is hugely popular in the global South. No teacher training needed, no handing out of devices, none of the usual challenges with edtech implementation. It just works.
With tools like this, how much do learners use AI to help with their homework versus do it? And how extensive is teacher usage of AI in the global South? Of course, this is not real edtech (as Prof Inge Molenaar notes, an LLM Is not an intelligent tutor), but I think the availability and ease of use of AI demand urgent attention.
AI in education, but is anyone learning?
We need evidence to understand better how children and teachers are using AI: productively and in service of learning, or uncritically and, increasingly, dependently. If the latter, the consequences will be dire indeed.
Countries with already low levels of education will be hollowed out further, not by AI cognitive augmentation but replacement. Instead of AI helping to learn, supporting student’s productive struggle as they do the hard work of gaining and applying knowledge, it’ll become a crutch. As one teen put it: “AI makes homework easier, but at what cost to our learning?”
Even if the access issue is solved, we may end up with AI-enabled classrooms and homes that are devoid of actual learning.
What to do?
As Mike notes well, to use AI productively, access to supportive people – engaged parents, private tutors, trained teachers – is critical. But we may need to add ‘adaptable systems’ to the mix. AI’s capabilities now in the hands of millions of children and teachers has forced a timely rethink of student assessment – something education systems have needed for years. At the systems level, new rules are needed for how to teach and learn well in an AI world.
Those that can adapt faster to these tech pressures will be on the right side of the digital divide. US President Trump recently issued an executive orderaimed at advancing AI education in K–12 schools and workforce training programs through federal initiatives. The order emphasizes promoting broad AI literacy, integrating AI appropriately into classrooms, training educators, and preparing the workforce for an AI-driven future.
China’s Ministry of Education issued guidelines to promote AI education in primary and secondary schools. Goals include cultivating students’ core competencies for adapting to an intelligent society, more innovative assessments and empowered teacher training. Some strict guidelines are included: At the primary school level, students are not permitted to use open-ended content generation tools independently.
Teachers are prohibited to rely directly on AI to answer students’ questions or provide consultation, and should avoid using AI-generated content for student assessment. China’s usual approach to AI regulation is sector specific and swift: It will be interesting to watch how these guidelines fare, what works and where improvements can be made.
Adapt or become AI dependent
AI is here to stay, inside and, even more so, outside the classroom. Countries may choose different approaches to the US or Chinese models, and hopefully we’ll see a range of programs that we can learn from before students and teachers become AI dependent. But they’ll need to move swiftly.
As the New Yorker article points out: roughly half of all undergrads in the US have never experienced college without easy access to generative AI. Perhaps the third digital divide is also between those that can adapt education systems quickest to a changing environment.
