AI-boost SCI: Pedagogical model, teacher competence and AI-resilient examination
The project uses best practice boosting the SCI faculty's AI competence via a tiered pedagogical model (for everyone/many/a few), a workshop series, SCI guidelines for students' AI use, an AI resilience matrix for examination and an experience bank seeded from three existing SCI pilot projects.
Project context
The rapid development of AI is changing the fundamental conditions for learning, teaching and examination. Generative AI tools have in a short time changed how students plan their studies, how learning activities are designed and how knowledge is made visible. Research shows that uncritical use of large language models can weaken learning, problem-solving skills and cognitive activation, while strategic integration of AI – as a critic, conversation partner or individual feedback support – can strengthen active learning. AI-resilient examination is described in the pedagogical literature as an intractable (wicked) problem that resists final solutions and requires coordinated schoolwork.
The SCI school's reform agenda 2026 identifies "high competence among teachers in AI tools and possibilities" as a central goal, with the ambition to achieve a high minimum level of AI competence via workshops at each department. This reform agenda needs to be operationalised and at the same time respond to KTH's priority "Education with AI" for the 2026 development project within the Future Education programme.
There are already three concrete AI pilot projects underway at SCI that can be used as reference cases and seed material for such a project. In addition, there are several external frameworks and models, such as the Chalmers AI strategy, the TU Eindhoven framework, the Elsevier AI checklist and the EU AI Act.
Purpose (outcome)
The purpose of the project is to ensure that faculty have the minimum necessary AI competence, that students develop documented AI knowledge, and that the examination is legally secure even when AI is a natural part of the learning environment. In addition, the project is a coordination project that brings together existing SCI initiatives and provides them with a common pedagogical, operational management and communication framework.
For KTH, the project aims to (i) reduce variance between courses, (ii) accelerate new teachers' introduction to AI pedagogy and (iii) create a scalable structure that other schools can reuse.
Note: Several measurable impact objectives exist – read more in the project directive below.
Project results (output)
The project will deliver the following concrete results:
- Pedagogical model for AI at SCI – with level division of learning objectives and abilities into three categories
- SCI guidelines for students' AI use – communication templates for course level with four disclosure levels
- AI resilience matrix and design guide for examination forms – concrete tools that help examiners choose examination forms.
- AI tool inventory with GDPR classification – list of evaluated tools marked Safe / Caution / Restricted
- Workshop series – four half-days for SCI faculty during autumn 26 and spring 27
- SCI Agentathon – a completed opportunity for teacher/student co-development; format documented for replication.
- Digital experience bank (community of practice) – SCI/KTH resource with documented cases, prompt library and evaluated tools.
- Three reference cases and 3–5 new pilot courses integrated and documented during spring 27.
- Final report with scaling proposals.
Time plan
Start date: 2026-03-01
End date: 2027-08-31
Project documentation
If you have a KTH ID, you can read the documents when logged in:
- Project directive (Swedish): Projektdirektiv_FrU26_2628-SCI_AI-lyft.pdf