Feedback Forensics: A Toolkit to Measure AI Personality

2025-09-30

AI Summary

Current AI model personality evaluation lacks publicly available, transparent quantitative tools, hindering detection of personality biases and potentially triggering model withdrawals or overfitting in feedback-based leaderboards. This paper introduces Feedback Forensics—a first-of-its-kind systematic framework for tracing AI personality evolution. It integrates AI annotators, LLM response comparison, and psychometric trait-scoring algorithms, leveraging mainstream feedback datasets (Chatbot Arena, MultiPref, PRISM) to quantitatively assess how human and AI feedback shape model personality. We release an open-source toolkit—including a Python API and an interactive web application—that enables reproducible, interpretable visualization of personality tendencies. Empirically, we uncover previously undocumented latent risks across major models, including sycophancy and consistency drift under varying feedback mechanisms. This work advances AI evaluation from opaque, black-box assessment toward explainable, auditable, and scientifically grounded personality analysis.

Technology Category

• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Humans and AI: Learning Human Values and Preferences
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Some traits making a "good" AI model are hard to describe upfront. For example, should responses be more polite or more casual? Such traits are sometimes summarized as model character or personality. Without a clear objective, conventional benchmarks based on automatic validation struggle to measure such traits. Evaluation methods using human feedback such as Chatbot Arena have emerged as a popular alternative. These methods infer "better" personality and other desirable traits implicitly by ranking multiple model responses relative to each other. Recent issues with model releases highlight limitations of these existing opaque evaluation approaches: a major model was rolled back over sycophantic personality issues, models were observed overfitting to such feedback-based leaderboards. Despite these known issues, limited public tooling exists to explicitly evaluate model personality. We introduce Feedback Forensics: an open-source toolkit to track AI personality changes, both those encouraged by human (or AI) feedback, and those exhibited across AI models trained and evaluated on such feedback. Leveraging AI annotators, our toolkit enables investigating personality via Python API and browser app. We demonstrate the toolkit's usefulness in two steps: (A) first we analyse the personality traits encouraged in popular human feedback datasets including Chatbot Arena, MultiPref and PRISM; and (B) then use our toolkit to analyse how much popular models exhibit such traits. We release (1) our Feedback Forensics toolkit alongside (2) a web app tracking AI personality in popular models and feedback datasets as well as (3) the underlying annotation data at https://github.com/rdnfn/feedback-forensics.

Problem

• Measuring AI personality traits without clear evaluation objectives
• Addressing limitations of opaque human feedback evaluation methods
• Providing tools to track personality changes across AI models

Innovation

• Open-source toolkit tracks AI personality changes
• Uses AI annotators for personality investigation
• Provides Python API and browser app interface