Originally published by the World Economic Forum.
The rise of generative artificial intelligence (AI) tools like ChatGPT is transforming the higher education landscape, sparking both excitement and concern. While many celebrate their potential to revolutionize how students learn, educators teach, and institutions function, others caution that they will quickly erode academic integrity by enabling wide-spread cheating and plagiarism.
While issues like these are valid and understandable, they overshadow a more critical dilemma: the educational paradigm that dominates global institutions is outdated and fundamentally unprepared for the age of AI.
A new model is urgently needed. One grounded in learning science and primarily focused on teaching students “how to think”' through the cultivation of 'durable skills' such as critical and creative thinking, ethical reasoning, and emotional intelligence. This paradigm not only future-proofs students by allowing them to do what AI cannot but also enables them to ethically and effectively utilize those tools and avoid being sidelined by them.
If we are to cultivate highly effective learners who can transfer their skills from the classroom to the real-world, then we need to ensure they actively recall and deliberately apply information across both time and context. They also need to receive ongoing, constructive feedback that enables them to target and address gaps in their understanding.
There are four reasons why the most popular and prevalent educational model (lecture + high-stakes exams) fails spectacularly to meet such conditions.
AI not only highlights the shortcomings of a traditional lecture-and-exam model but actively undermines its value. Students will rightly question why they should attend lectures when AI can interpret, visualize, and summarize complex information for them whenever and however they want, and in ways tailored to their readiness level and needs. The single grade they receive on an exam or essay cannot match the highly personalized (real-time) formative feedback AI tools can offer them. And if their value upon graduation lies merely in their ability to recall specialized knowledge, then they will soon realize that AI tools can duplicate that skillset, at much lower cost and with increased efficiency, making them surplus to requirement.
Clearly a new model is needed. One which seeks to supercharge teaching, learning, and assessment using insights from the science of learning. That students not only need to know “what to think” but more importantly “how to think”. And one which prepares them to ethically, effectively, and critically use, as well as make decisions based on AI tools and their output. In short, a model which endows them with AI-resilient “durable skills”.
A skill-based model structures the learning journey around skills mastery, with foundational skills serving as the bedrock upon which more complex skills are built within and between courses. Although students gain specialized knowledge (perishable skills) and a familiarity with generalized frameworks (semi-durable skills), both serve as a method to continually hone their durable skills (rather than endpoints unto themselves).
Such an approach discards lectures for a “flipped-classroom” where skills are independently acquired at home and then applied to real-world problems during class using active learning techniques like Socratic discussion and simulation. High-stakes exams and essays are similarly replaced with experiential assignments based on real-world challenges and connected to local partners. These not only allow students to apply their skills to authentic issues but to be evaluated based on their capacity to iteratively reflect, revise, and improve their thinking based on the dynamic challenges they are facing.
A skills-based model is far better positioned for an AI era than traditional models because it is:
AI tools can - and should - be leveraged by academic leaders, educations, and students in a skill-based model. For instance:
1. Curriculum Design & Skills Mapping: Academic leaders can use AI to (a) analyze market trends, job descriptions, and industry demands to identify the skills they want their students to acquire; (b) organize those skills into a hierarchical taxonomy, and (c) suggest how to sequence skills in order to build a scaffolded curriculum.
2. Content Generation: Educators can use AI to generate creative active learning exercises that directly target skills being trained in their course. AI can also help generate ideas for experiential assignments by pulling from databases of current industry challenges or on-going community issues, thereby allowing students to apply those same skills to complex real-world questions.
3. Adaptive Learning & Feedback: Educators can also use real-time data and feedback from AI to adapt their instructional methods on the fly, and to help automate grading.
Elsewhere, students can use those same tools to dynamically adjust the difficulty or focus of learning materials, based on how well they are mastering a skill or concept, and to receive continuous real-time formative feedback based on their performance.
4. Performance Metrics: AI can help academic leaders monitor the effectiveness of their new skills-based model by tracking key performance indicators, including student engagement, skill acquisition rates, and their relation to graduate success metrics. This information can be used to provide feedback on the durability of the skills being trained and to direct ongoing curriculum refinement.
For an in-depth treatment of the integration of AI in Higher Education, read the full white paper.
