Master’s Thesis in Deconstructing Reasoning Traces in LLMs
About The Position
As Sweden's national center for applied AI, AI Sweden is looking for an ambitious Master’s thesis student to join them, in close collaboration with the Stef SM Lab at Chalmers University of Technology. The project focuses on understanding the nature of reasoning traces in modern AI systems. These traces, which involve chains of reasoning, self-checking, and backtracking, are crucial for recent AI advancements but are not well understood. The research will explore whether written reasoning is the actual computation or a narrative about it, the relationship between reasoning length and reliability/confidence, and the potential of reasoning traces as an attack surface or security control. The student will work on open research questions in a frontier area of AI, with access to substantial compute resources. Projects can be theoretical, computational, or a combination, involving designing controlled experiments, running them on open-weight reasoning models, and measuring results. The thesis can be completed individually or in pairs.
Requirements
- Ongoing Master’s studies in Computer Science, Data Science, Engineering Physics, Complex Adaptive Systems, Machine Learning, or similar.
- Comfortable with Python and deep learning frameworks.
- Comfortable with the idea that an experiment might yield unexpected results.
- Curious, independent, and self-driven.
Nice To Haves
- Prior experience with language models.
- Prior experience with reinforcement learning.
- Genuine curiosity and interest in AI safety and security.
Responsibilities
- Design controlled experiments to understand the reasoning phenomenon in LLMs.
- Run experiments on open-weight reasoning models.
- Measure and analyze experimental results.
- Potentially focus on adversarial robustness for reasoning traces, honesty of reasoning, distinguishing reasoning from recall, or visibility of cognitive processes.
- Collaborate with AI scientists and researchers at AI Sweden and Chalmers University of Technology.
Benefits
- Opportunity to work at the frontier of AI research.
- Access to substantial compute resources.
- Work on genuinely open research questions with a chance for impactful results.
- Collaborate with leading AI scientists and professionals.
- A place to grow and an environment for personal development and achievements.
- Work in a unique position combining governmental influence and startup agility.
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Job Type
Full-time
Career Level
Entry Level
