analiza aiqbench
This paper identifies a critical gap in current eXplainable AI (XAI) frameworks, which treat models as static entities rather than dynamic systems undergoing behavioral shifts via interventions like fine-tuning or reinforcement learning. It proposes Comparative XAI (XAI$_\Delta$), a new paradigm designed to explicitly explain the causal transitions between model checkpoints, addressing urgent governance needs under regulations such as the EU AI Act. By focusing on the functional transformation itself, this approach offers a principled method for auditing and documenting substantial system modifications in large language models.
Comparing Explanations is Not Enough, Explain the Change: New Standards are Needed to Explain Behavioral Shifts in Large Language Models
arXiv:2602.02304v3 Announce Type: replace Abstract: Large-scale foundation models exhibit behavioral shifts when subjected to interventions such as scaling, fine-tuning, reinforcement learning with human feedback, or in-context learning. Current explainability methods are structurally ill-suited to explain these shifts, because they either treat models as static objects, as traditional eXplainable AI (XAI) approaches do, or merely compare independent explanations across different checkpoints of a model. As a result, these approaches fail to explain the functional transition between two model instances in which a certain behavior has shifted following an intervention. This gap creates significant governance risks across jurisdictions including the EU AI Act, US state legislation, and Chinese AI regulations, which require documenting causal chains for substantial system modifications. This position paper argues that explaining behavioral shifts in large language models requires a principled approach that treats the shift itself as the primary object of explanation: namely, one that explains how and why an intervention transforms a reference model into an updated model with different behavior. To support this claim, we introduce Comparative XAI (XAI$_\Delta$), a novel XAI paradigm aimed at explaining the difference between two model checkpoints where a behavior has shifted, together with a set of desiderata specifying what XAI$_\Delta$ explainers and explanations must satisfy, including comparability, validity, actionability, and monitoring, with the goal of grounding model auditing in explicit, measurable requirements. Finally, we provide preliminary evidence suggesting the need for XAI$_\Delta$ in practice through illustrative experiments, compiling the resulting findings into a transition report directly usable for governance and incident documentation.
Abstract:Large-scale foundation models exhibit behavioral shifts when subjected to interventions such as scaling, fine-tuning, reinforcement learning with human feedback, or in-context learning. Current explainability methods are structurally ill-suited to explain these shifts, because they either treat models as static objects, as traditional eXplainable AI (XAI) approaches do, or merely compare independent explanations across different checkpoints of a model. As a result, these approaches fail to explain the functional transition between two model instances in which a certain behavior has shifted following an intervention. This gap creates significant governance risks across jurisdictions including the EU AI Act, US state legislation, and Chinese AI regulations, which require documenting causal chains for substantial system modifications. This position paper argues that explaining behavioral shifts in large language models requires a principled approach that treats the shift itself as the primary object of explanation: namely, one that explains how and why an intervention transforms a reference model into an updated model with different behavior. To support this claim, we introduce Comparative XAI (XAI$_\Delta$), a novel XAI paradigm aimed at explaining the difference between two model checkpoints where a behavior has shifted, together with a set of desiderata specifying what XAI$_\Delta$ explainers and explanations must satisfy, including comparability, validity, actionability, and monitoring, with the goal of grounding model auditing in explicit, measurable requirements. Finally, we provide preliminary evidence suggesting the need for XAI$_\Delta$ in practice through illustrative experiments, compiling the resulting findings into a transition report directly usable for governance and incident documentation.
Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2602.02304 [cs.AI]
(or arXiv:2602.02304v3 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2602.02304 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 342 cuvinte
Subiecte conexe
Untrusted advice for AI control: Short, strong advice significantly uplifts weak LLMs
LessWrong · 27 iul. 2026
The current bottleneck is political will, not research
AI Alignment Forum · 11 iul. 2026
[AINews] Opus 5.5 is good at explainer videos
Latent Space · 29 sept. 2026
[AINews] The Future of Latent Space
Latent Space · 25 sept. 2026