Enhancing Explainability in Large Language Models Through Belief Change: A Simulation-Based Approach

Authorea (Authorea), Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 20, 2024

Artificial intelligence systems, particularly those deployed in high-stakes environments, require a high degree of transparency and explainability to ensure that their decisions can be understood trusted. Traditional approaches enhancing often rely on post-hoc methods fail fully capture the internal reasoning processes complex models. In this research, novel integration Belief Change Theory was employed address challenge, offering systematic framework for belief revision directly influences decisionmaking process model. The proposed methodology implemented Llama model, which modified incorporate mechanisms capable handling contradictory information generating coherent explanations. Through series simulations, model demonstrated significant improvements consistency, accuracy, overall explainability, outperforming traditional models lack integrated management systems. findings highlight potential not only enhance AI systems but also provide foundation more dynamic interactive forms interpretability. research opens new avenues development are both powerful accountable, paving way adoption critical decision-making contexts.

Language: Английский

Automated Early Detection of Misinformation on Social Media: A Large Language Model Approach with High-Volume Facebook Data

Published: Aug. 13, 2024

Social media platforms have become a primary conduit for the rapid dissemination of information, where unchecked spread misinformation poses significant threat to public discourse and societal well-being. Introducing an innovative approach that leverages advanced capabilities fine-tuned ChatGPT model, this research addresses urgent need scalable accurate methods detect in real-time across vast digital landscapes. The model was meticulously evaluated through series experiments demonstrated its superior performance identifying misleading content, particularly when compared traditional machine learning classifiers earlier versions language models. integration comprehensive preprocessing techniques, alongside refined confidence thresholds post-processing rules, enhanced model's ability process complex diverse datasets, resulting highly reliable predictions. findings underscore potential significantly mitigate misinformation, offering solution capable operating effectively fast-paced environment social media. By advancing field detection, study provides critical insights tools can be applied both practical domain content moderation, contributing more informed resilient society.

Language: Английский