AI Tools Strengthen New Nuclear Energy Research and Development

EXPLORE

AI Tools Strengthen New Nuclear Energy Research and Development

Author: Professor Anasse Bari & Professor David J. Nagel

August 25, 2024, New York City - The world urgently needs new and expanded clean energy sources due to the growing global population, the increasing per capita use of energy as countries develop, and the many clear and increasing effects of climate change. The rise in global sea levels is particularly worrisome.

Nuclear energy is a viable option for current and future power needs as it does not release greenhouse gases. Hence, there is significant interest in fission energy and various fusion approaches, both receiving substantial funding. Another potential type of practical nuclear energy is LENR, which stands for Low-Energy Nuclear Reaction, a phenomenon that is also referred to as Solid-State Fusion (SSF) or Cold Fusion. The United States Department of Energy recently allocated $10 million to research whether LENR could serve as a potentially transformative carbon-free energy source.

Professor David Nagel, with over 60 years of experience as a scientist at the United States Naval Research Laboratory and the George Washington University, has been collaborating with Anasse Bari, a Professor of Computer Science at New York University (NYU), who leads the Predictive Analytics and Artificial Intelligence (AI), research laboratory at NYU Courant. Together, they are leading a team at NYU to design and deploy AI tools to advance the research and commercialization of this field.

The NYU Program on AI for LENR Research, sponsored by the Anthropocene Institute, has two main objectives: to apply existing AI algorithms to structure and analyze the extensive body of literature on LENR, to uncover actionable insights that are otherwise unobtainable; and to design and deploy new AI tools to support and accelerate ongoing and future research, development, and commercialization of LENR.

The extensive LENR research literature, consisting of a vast and unstructured data collection, is too large for individuals to assimilate. With over 5,000 reports, publications, presentations, and articles scattered across various sources on the Internet, the field presents significant intellectual and operational challenges. AI's advanced capabilities can significantly benefit the LENR community by advancing the field and supporting research efforts. The use of modern AI tools will greatly benefit both the understanding and commercial development of LENR.

On April 10, 2024, Nagel delivered a lecture hosted by New York University. He addressed around two hundred attendees on the Status, Momentum, and Potential of Low Energy Nuclear Reactions. Nagel reviewed the substantial progress and serious promise of LENR and cited some of the challenges that must be overcome prior to its commercialization. They range from thorough scientific understanding to developing effective and durable active materials to engineering commercial generators.

Prof. Nagel's lecture at NYU is here: https://youtu.be/vKxGNhWEShU

A DataHub for LENR’s Literature

The challenge at hand involves over 35 years of LENR experimental reports and publications that remain unstructured and not readily usable for AI applications. The information is scattered across various sources on the internet, making it difficult to utilize effectively. The team has collected over 5,000 publicly available LENR-related publications to address this and created a structured data hub. The centralized hub now houses structured data ready for machine learning. This enhancement significantly improves accessibility and utility for research and AI-driven analysis. The team has released the first version of their report on the data, based on a comprehensive study presented in Japan at the 2024 9th IEEE International Conference on Big Data Analytics (ICBDA). You can access the study here: IEEE Xplore (DOI: 10.1109/ICBDA61153.2024.10607191). Additionally, an experimental dashboard showcasing the LENR data and tools is available here: LENR Dashboard. Several AI tools are available on the Dashboard under the EXPLORE tab.

Exploring Artificial Intelligence Techniques to Research Low-Energy Nuclear Reactions

Leveraging modern AI capabilities, particularly advanced natural language processing techniques, the team has just released a study published in the journal Frontiers in Artificial Intelligence. This study summarizes the results of employing advanced embedding models and topic modeling techniques to analyze and understand the extensive body of LENR research. By utilizing tools like Latent Dirichlet Allocation (LDA), BERTopic, and Top2Vec, researchers have uncovered relationships and trends within the LENR field. The study also introduces LENRsim, an experimental machine learning tool designed to identify similar LENR studies. These advancements in AI-driven analyses offer new insights and pathways for researchers, significantly contributing to the future of LENR research and the development of clean energy solutions. The study was published in Frontiers in Artificial Intelligence on 22 August 2024, Section: Machine Learning and Artificial Intelligence, Volume 7 - 2024 and can be accessible at: https://doi.org/10.3389/frai.2024.1401782

An Experimental LENR Chatbot

The growing popularity of Large Language Models (LLMs) for chatbots highlights the need for a specialized chatbot focused on Low Energy Nuclear Reactions for scientific research. The team has designed and deployed a preliminary experimental LENR chatbot that exclusively uses pre-processed documents from the LENR DataHub. The data is provided to the chatbot through both automated and manual processes, ensuring it only accesses LENR-related information to minimize hallucinations. Currently, in the testing phase, the team is continuously refining the chatbot and researching new algorithms to enhance its capabilities. The chatbot can also be accessed on the SSF website https://solidstatefusion.org/lenrbot/

The team, supervised by Bari and Nagel, includes researchers from New York University: Sneha Singh, Yvonne Wu, Tanya Pushkin Garg, Benjamin Kang, Emos Ker, Charles Wang, Dongjoo Lee, Adelina Simpson, Suryavardan Suresh, Yifei Xu, Shreya Guda, Anway Agte, Sathvika Bhagyalakshmi Mahesh, Harshini Raju, and Ian Liao.

Upcoming Presentation

Professors Nagel and Bari will be delivering a presentation titled "Utilizing Artificial Intelligence for Advancing LENR: Mining and Mastering the LENR Literature" on September 3rd, 2024 at the 16th International Workshop of Anomalies in Hydrogen Loaded Metals (IWAHLM-16). The event, hosted by the International Society for Condensed Matter Nuclear Science (ISCMNS) in collaboration with the Société Française de la Science Nucléaire dans la Matière Condensée (SFSNMC), will take place from Sunday evening, September 1, through Wednesday, September 4, 2024, both in Strasbourg, France and online.

For more information, visit: IWAHLM-16 Workshop

EXPLORE