Artificial Intelligence

Description

Offer Description: As part of the research project, you will contribute to the task entitled “Interpreting the Internal Mechanisms of Models,” which aims to analyse and gain an in-depth understanding of how AI models work internally by establishing explicit connections between their internal structures, including weights, activations and gradients, and human-interpretable concepts. The objective is to move beyond purely behavioural interpretability and develop a mechanistic and causal understanding, which is essential to ensuring the transparency and reliability of AI systems. Your work will focus on mapping and understanding the internal representations of AI models by identifying alignments across modalities and disentangling latent concepts. You will also analyse learning dynamics and design architectures that promote both performance and interpretability. Finally, you will apply mechanistic understanding to advanced reasoning by making implicit reasoning processes understandable and reducing hallucinations. Requirements: Research Field Communication sciences » Science communication Education Level PhD or equivalent Skills/Qualifications PhD in Computer Science, AI, or Machine Learning. Strong skills in deep learning, explainable AI, Python, and PyTorch or equivalent frameworks. Specific Requirements AI, or Machine Learning. Strong skills in deep learning, explainable AI, Python, and PyTorch or equivalent frameworks. Languages ENGLISH Level Excellent

Sectors

Source

Source: euraxess.ec.europa.eu