Together against the machine

Mapped principles

The project is mapped to the following framework principles:

P3. Active student-centred learning

P4. Assessment and examination for learning  

Definitions of the principles

Contact

Lars Nordström professor

larsno@kth.se , 087906830

Profile

Project number 

This project is implemented at the School of Electrical Engineering and Computer Science (EECS) within the Future Education programme (project no. 2619-EECS) 

Project context 

Programming projects as an assessment format are common in KTH’s engineering programmes. With the emergence of powerful AI assistants for programming, the conditions for using functioning source code as proof of assessment have fundamentally changed. Students are now able to generate technically correct code without having developed the necessary understanding of design, quality and problem analysis.  

Fundamentally, this is not only an assessment problem, but part of a development that future engineers, not least those who are not computer scientists by background, must relate to. Students need to practise and be assessed on how to relate to the fact that the machine can program itself. This includes when to trust AI-generated code, how to review it, and how to work together with AI agents rather than simply using them as black boxes. 

The project is a new initiative within Future Education 2026, but connects to ongoing work at EECS on project-based courses and industry collaboration. It has a clear link to the industry partner Svenska kraftnät, where questions about AI use in systems development are highly relevant.  

Purpose (outcome) 

The purpose of the project is to prepare students for a working life in which they will work together with AI agents, and in some cases compete with them. Future engineers, not least in subjects outside computer science, must be educated to take a responsible approach to the use of AI, especially in systems development and programming. 

The intended outcomes are that: 

• Students who have completed the course can critically review AI-generated code and justify design choices in relation to quality requirements 
• The assessment measures student’s learning rather than the performance of the tools, even when AI assistants are used 
• The method can be adopted by other course coordinators at EECS and other KTH schools. 

Measurement will be carried out through (a) comparison of qualitative assessments of student performance before and after implementation, (b) student surveys on perceived understanding and independence, and (c) the number of courses and course coordinators who make use of or adopt the method during and after the end of the project.  

Project results (output) 

The main output of the project is a method for continuous assessment of students working in groups in problem-based learning, where the aim is to move from an overall functional requirement, through analysis and design, to programming to solve the problem. The method consists of four integrated parts: 

• A seminar series on ways of working with AI assistants (preparation material, guide to run it, and evaluation form). 
• Knowledge tests on software quality (question bank with marking instructions). 
• A format for student-led peer review sessions throughout the project, including preparation templates and documentation requirements.​​​​​
• An assessment format called “human versus machine” opposition, with instructions and assessment criteria. 

In addition to the method, the project outputs are: 

• Pilot implementation of the method in a selected course during autumn term 2026 (parts of the course) and spring term 2027, with continuous documentation of implementation, lessons learned, and student response. 
• Implementation in a live course offering of a revised version during autumn term 2027, with evaluation and comparison with previous assessment formats. 
• A final report describing the design, implementation, evaluation, and recommendations for continued use of the method. 

^{*The method is deliberately designed so that, with contextual adaptations, it can be used by other examiners (in other subjects and schools withing KTH) without the whole format needing to be rebuilt.}

Time plan 

Start date: 2026-06-01
End date: 2027-12-17 

Project documentation 

If you have a KTH ID, you can read the documents when logged in:

• Poster: The project’s first poster (in Swedish) will be available at the ”Storträffen” meetup autumn 2026
• Project directive (Swedish): Projektdirektiv_FrU26_2620-EECS_AI-förstärkt.pdf