Mid Sweden University has launched a cross-faculty project that explores artificial intelligence as a student-centered learning tool. Rooted in the university’s general policy for generative AI use, the project sets out to answer a common question: How can AI meaningfully enhance learning rather than replace it? Five discipline-specific pilots were designed to probe this question while sharing the results with all subjects at the university.

During this project, the following research questions were therefore investigated:
1) In what ways can AI applications scaffold deeper learning and engagement across diverse subjects?
2) Which strategies maximize student throughput while respecting institutional AI guidelines?

Each pilot adopted its own method and had different starting points and end goals, but they all have been gathered here in this article for comparative analysis. In order to gather lessons learnt from all the different pilots and create a good comprehensive overview. In detail, the pilots and their subjects were:

Computer Engineering:
Using generative-AI as catalysts for learning. By inserting customized generated contents after each lecture to encourage student reflection and peer discussion. All in order to create an improved study pace, course retention, and course throughput.

Law:
In an introductory course to law, students in human resources made a comparison between three different legal AI-tools (ChatGPT, Copilot and Blendow Lexnova). The students were required to ask the tools two specific questions regarding labor law and one optional question. Thereafter the students evaluated, compared and discussed the answers based on, for instance, the traffic-light model.

Mathematics:
AI supported essay writing in mathematics courses. The lecturer introduces a concrete, relatively advanced engineering problem (stabilizing the reversed pendulum) and gives hints how AI is used to solve it. The participants write a short essay freely related to the topic and are encouraged to use various AI tools and to summarize how they have worked.

Psychology:
AI-driven case simulations illustrating treatment processes for disorders ranging from anxiety to depression, allowing students to contrast cognitive, behavioral and psychodynamic approaches.

Teacher Education:
A 15-credit course was redesigned to provide teacher students with theoretical understanding and practical experience in applying AI to educational contexts. Guided by the TPACK, SAMR, and Traffic Light frameworks, students engaged with a range of AI tools in tasks involving translation, lesson design, and ethical reflection. Thus, improving metacognition, critical thinking, and inclusive teaching practices, demonstrating how AI can be integrated into our teacher education.

Our initial results and evidence confirm that when combining robust pedagogical models and transparent ethics, generative AI we can amplify rather than undermine the learning process. Our results varied in the different pilots, but in general the results can be summarized as follows: Across all subjects, reflection on how, when, and for what purpose to use AI emerged is the key point in preventing AI from becoming a “shortcut” and instead positioning it as a catalyst for deeper learning. In future work we will further explore long-term retention and expand our development resources to scale up our pilots to become successful tactics university-wide.

Keywords: Artificial intelligence, Student engagement, Interdisciplinary pilots, Higher education.