Can we critically think about the future in the age of generative AI? 

In this blog, Luciana Blaha, Assistant Professor in the School of Social Sciences (Edinburgh), poses three challenges and three potential solutions to help staff and students think critically in the age of GenAI.   

We are surrounded by impressive headlines every day, from the graduate labour market contracting by 40% (Financial Times, 2025) to rushing towards building AI agents, all while trying to anticipate tomorrow’s world of work for our learners.  

Generative AI is changing how students learn, how graduates enter work, and how universities think about knowledge. For lecturers, the question is no longer whether AI matters, but how to respond in ways that strengthen learning rather than reduce it to either fear or hype. So what does academic research tell us about the current higher education teaching approaches to employability? I propose 3 key challenges, and 3 potential solutions that we can offer as educators in this AI-enhanced world of work.  

Crystallised and Fluid Intelligence 

The world of work was built on a somewhat stable set of assumptions, which centred thinking roughly around organisations’ financial success. Much of education has traditionally rewarded accurate recall, repetition, and compliance with established knowledge in a ‘crystallised’ intelligence framework (Cady et al, 2024). These are still important, but they are also the tasks most easily supported by generative AI. A key set of recommendations therefore focuses on nurturing the ‘fluid’, uniquely human types of intelligence which not only resist replacement by automation and AI, but also foster emotional wellbeing, empathy, complex problem-solving, and collaboration (ibid). Moreover, as Andel et al, and Kriger et al (cited in Cady et al, 2024) show, cognitively stimulating occupations like specialised, scientific, analytical and strategic work have been associated with less cognitive decline, dementia, and an increased ability to counteract aging-related challenges.  

Activities to promote this include: promoting experiential learning, redesigning jobs and processes (in line with reviewing expertise to include complexity and creativity (Faraj et al, 2018)), engaging in participatory communication (Forde-Stiegler et al, 2024), and stimulating cognitive flexibility (Cady et al, 2024).  

Cognitive Load 

Students are learning in environments saturated with information, distraction, and digital tools. AI can help reduce some of that burden (by supporting first drafts or helping students navigate unfamiliar content), but it can also create more noise, and the temptation to rely on shortcuts. We have limited working memory. It is reduced by factors such as task complexity, the novelty of the information, and our prior knowledge around the area, which impact on how ‘heavy’ we perceive the task to be cognitively. While findings vary somewhat between newer types of AI like Generative AI and Machine Learning, research shows that the relationship between human learning and development and AI can be influenced by educators, both positively and negatively (Gkintoni et al, 2025). The ability to foster positive outcomes depends on acknowledging the cognitive load while reducing routines, minimising distractions, and stimulating cognitive flexibility through questioning.  

A practical example of this is provided by Tiwari and Bhagat (2024), where humans used LLMs to reduce the general content they needed to sift through before finding the meaningful knowledge to their task, while still maintaining awareness of it through prior learning. This process led to both cognitive development, and decreased psychological stress.  

Critical Thinking 

Critical Thinking (CT) skills (analysis, evaluation, and inference) refer to the higher-order, cognitive, ‘task-based’ processes (Dwyer et al. 2014) which we associated earlier with cognitive flexibility and fluid intelligence. Yet current Critical Thinking research places these skills as a combination of ‘dispositions’, mixing attitudes, motivations, and habits which are highly dependant on one another, and therefore must be encouraged  jointly by educators, at home (parents), in institutions and the learners themselves. In AI-rich classrooms, students may treat AI-generated content as authoritative unless they are explicitly taught to test it, compare it, and reflect on its limits, but they must also take ownership of their life-long learning. When faced with AI, environments fostering a separation between potential employment security and personal development are more likely to encourage literacy and new technology adoption (Blaha, 2021). So how can we maintain motivation, develop habits, while also encouraging inquisitive attitudes? 

Case-Based Learning (CBL) and Problem-Based Learning (PBL) are two approaches that can introduce the complexities of the real-world into learning to link learned information with practice. CBL looks at specific cases where examples can be given about a specific situation (Apollo missions, GenAI incidents) (Pinto, 2022). PBL builds on CBL by having many solutions, and requiring students to craft an answer to the best of their ability critically and without the instructor, for example looking at wicked problems (climate change, labour market changes).  

Examples of CBL / PBL approaches to AI include:  

• students could analyse a realistic AI-related case,  

• compare their own judgement with an AI-generated response, and critique the differences in class (CBL),  

• or work in groups on an open-ended challenge, such as designing a responsible way to use AI in teaching or assessment. They may use AI to explore options but must justify their final solution in terms of ethics, feasibility, and academic integrity (PBL). 

To prepare students for AI-shaped futures, we should create low-risk opportunities to experiment, question outputs, and reflect on consequences. Used this way, AI becomes not a shortcut, but a stimulus for judgement, adaptability, and lifelong learning grounded in ethical practice. 

We have to experience AI and be able to safely play, and reflect on it in order to make effective use of it. This stimulates cognitive flexibility by helping us learn deeply through experience, develop our cognitive flexibility, and reinforce it through critical thinking.  

To piece it all together once more, have a look at our TEDx Talk. 

References 

Blaha, L. (2021). Reconceptualizing Intelligent Automation: A Scottish Case Study. Doctoral Thesis, University of Aberdeen. 

Cady, S. H., Willing, J. G., & Cady, D. A. (2024). The AI imperative: On becoming quintessentially human. The Journal of Applied Behavioral Science, 60(4), 721-731. 

Dwyer, C. P., Hogan, M. J., & Stewart, I. (2014). An integrated critical thinking framework for the 21st century. Thinking skills and Creativity, 12, 43-52. 

Gkintoni, E., Antonopoulou, H., Sortwell, A., & Halkiopoulos, C. (2025). Challenging cognitive load theory: The role of educational neuroscience and artificial intelligence in redefining learning efficacy. Brain Sciences, 15(2), 203. 

Pinto, M. M. C. G. (2022). Problem-Based Learning: Concept and Method. Revista Gênero e Interdisciplinaridade, 3(01). 

Tiwari, A. S., & Bhagat, K. K. (2024). Comparative analysis of augmented reality in engineering drawing course: Assessing spatial visualization and cognitive load with marker-based, markerless, and web-based approaches. Australasian Journal of Educational Technology, 40(6), 19-36. 

Image credits:

Header image: Photo, ‘Person sitting in a maze’ from Microsoft stock images