Discover more from Dr Phil's Newsletter, Powered by DOMS™️ AI
AI + the Education System
The key to relevance, value & excellence?
Earlier today, I gave a keynote to a wonderful and energetic audience of academics and innovators at the University of Copenhagen. Together, we explored the risks of adopting (and not adopting) AI in formal education.
Here’s the lowdown:
Introduction: Education’s Identity Crisis
There has been much discussion in recent months about the risks associated with the rise of generative AI for education, with most of the discussion centring around the challenge that ChatGPT poses to academic integrity. It goes like this:
If AI can write an essay better than a student, what does this mean for an education system which assesses both its students and, to a large extent, its own success on the quality of written assignments?
Much less work has been done on exploring the negative – even existential – consequences that might stem from not embracing AI in education.
So let’s take a look at an ongoing discussion from a different angle.
Save the Assignment!
For many education institutions, the reaction to the rise of gen AI has been to increase monitoring and surveillance of students. For some, this has meant returning to in-person essay writing; literally requiring students to write essays under the watchful surveillance of educators.
Others have attempted to solve the same problem with a new flavour of plagiarism technology - so called ““AI detection” tools like GPTZero and Turnitin AI which, ironically, use AI algorithms to to detect assignments which are written by AI rather than humans. Or so they claim….
We now have a tonne of research to show that perfect AI detection isn’t possible; machines are not great at differentiating between human and AI generated text and are also deeply biased in their assessment. As one paper concluded, “GPT detectors unintentionally penalise both non-native English writers & all writers with constrained linguistic expressions”.
Meanwhile, the growth of AI-detection technology has spawned the creation of another new technology: so-called “Anti-detection” technologies like Undetectable which are built (again, using AI) to enable students to “humanize” AI-generated content and “fool the AI detectors”. Which, by the way, they do pretty well.
The headline here is that in an attempt to “save the assignment” and - perhaps - dig in our heels and preserve education as we know it, we’re caught in a bizarre farce which is both powered by and designed to repress AI. Yikes!
The campaign to “save the written assignment” (and, perhaps, our identities as educators and institutions) hasn’t gone so well so far. So what might a different, more productive approach look like?
The Purpose of Education
To answer this question, I asked another question: what is the true purpose of education?
Having a read a tonne of discussion about this, including definitions by UNESCO and the Danish Government, formal education is broadly considered to have two purposes or - some might say - responsibilities:
To develop knowledge and skills which drive real-world economic, social, and cultural growth.
To ensure the best possible quality educational experiences for all students.
Can AI help to help deliver on these goals? The answer is yes.
Goal 1: To develop knowledge and skills which drive real-world economic, social, and cultural growth.
In its Future of Jobs report 2023, the World Economic Forum reported that by 2025 AI will generate a net gain of 58 new million jobs, almost all of which will require some element of AI and machine learning expertise.
By 2030, AI's contribution to global GDP is estimated to reach $15.7 trillion, a 26% increase from today, with increased productivity powered by AI accounting for about 40% of this growth.
The headline here is that if formal education is to deliver on its promise to empower students to take an active part in society and contribute to its further development, they ned to develop knowledge, competence and confidence in how to build and use AI as part of their formal education.
Goal 2: To ensure the best possible quality educational experiences for all students.
As things stand, the predominant “chalk and talk” approach to teaching and learning in higher education is not optimised for the achievement of all students.
Over 30 years of research has shown that this “knowledge transfer” model enables students to recall enough to write an assignment or pass an exam, but it has little measurable impact on their skills and limited value beyond the educational setting, out in the real world (hence the need for so called “graduate traineeships” and the complaints of employers about the shortcomings of university education).
The headline here is that if formal education is failing to deliver on its quality promise, and that embracing AI might help it to a better job. How? Here are just two examples:
Needs Analysis: Great quality, equitable learning experiences start with data-driven needs analysis: i.e. a reliable understanding of who our students are, their Zone of Proximal Development (i.e. what they already know) and what motivates them. AI can help with this in a number of ways:
Data generation: it is possible to use AI survey and analysis to write, circulate, gather and analyse the results from learner needs analyses in seconds. Discovery work that used to take me two weeks of preparation, admin and analysis can now be completed in ~1 hr in total. The impact of this is potentially transformative.
Data analysis: AI is essentially a data analysis machine. It’s now possible to use AI data analysis tools to rapidly analyse, theme and report on years of student data extracted from surveys, LMSs and other sources. This data holds a wealth of untapped, hidden information about the value and impact of our teaching practices - thanks to AI it’s now possible to tap into it.
Instructional Design: Great quality learning experiences require teachers to keep up to date with what to teach and how to teach it. This is much harder than it sounds: a complex process of sourcing, reading and translating ever-changing learning science research into actionable principle for learning design. AI can help with this in a number of ways:
Rapid research review: AI-powered tools like Elicit are able to summarise up to date research on the most effective strategies for teaching most topics to many different learner types, e.g. “What does the research tell us about how to teach leadership online to junior marketing professionals?”. Combined with tools like ChatGPT, this makes it faster and easier than ever to review and optimise how we teach: upload your existing approach, paste in Elicit’s advice on best-fit instructional strategies and ask ChatGPT to optimise your instructional design for impact.
Rapid differentiation: AI tools also make it possible to design differentiated content. Try using your needs analysis data and an AI-powered tools like Elicit to generate research on the most effective strategies for teaching your topic to your specific learner type(s), e.g. “What does the research tell us about how to teach leadership online to junior marketing professionals with ADHD/no prior experience etc?”. Combined with tools like ChatGPT, this makes it faster and easier than ever to create multiple, differentiated pathways to the same outcomes.
Conclusion: A Crossroads for Formal Education
The conclusion we came to today is that formal education stands at a crossroads: it can either take action to shore-up what it already does and maintain established methods of designing, delivering and assessing teaching and learning. Or, it can leverage the power of emerging technologies like AI to innovate and shore-up its relevance, value and excellence in the years ahead.
Ultimately, the question of whether AI will disrupt education is less a question about technology and more a question about humans and institutions. The impact that AI will have on formal education will ultimately be dictated not by the existence of the technology (which is here, ready and waiting) but by the appetite among educators & education institutions for change and on their ability to execute it.
PS: If you design learning experiences and want to get hands on and experiment with AI supported by me, you can apply for a place on an upcoming cohort of my AI-Powered Learning Science Bootcamp here.