Department
Computer Science and Cybersecurity
Document Type
Poster
Abstract
Hardware-Level Approaches to Reducing AI Hallucinations in Large Language Models Abstract: Large Language Models (LLMs) are widely used today for generating text, answering questions, and assisting with many tasks. However, they often produce hallucinations, where the model generates information that is incorrect or not based on real data. Most existing solutions try to fix this issue using software techniques such as prompt engineering or retrieval systems. This research looks at a different perspective by examining hardware-level solutions that may help reduce hallucinations during model inference. Several peer-reviewed studies are analyzed that propose improvements in AI accelerators, memory management, and hardware-supported verification mechanisms. These approaches aim to make the computation process more stable and reduce the chances of errors spreading through neural networks. The findings suggest that combining better hardware design with AI systems could improve the reliability of large language models and help create more trustworthy AI systems in the future. Keywords: AI hallucination, hardware acceleration, LLM reliability, neural networks, AI architecture
Publication Date
Spring 4-9-2026
Recommended Citation
Remember to check citations for accuracy before including them in your work.
Virani, Neha, "Hardware-Level Solutions for AI Hallucinations in Large Language Models" (2026). Student Scholarship. 48.
https://metroworks.metrostate.edu/student-scholarship/48
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Comments
Spring 2026: Student Research Conference
Distinguished Presenter Award, Neha Virani