Lucerna Veritas sits at the intersection of science, philosophy, artificial intelligence, and art. Rather than beginning from disciplinary boundaries, it follows questions wherever they lead—from relational intelligence with large language models to atomic physics, from scientific measurement to public trust, from physical experiments to the nature of evidence itself.
The work is guided by three principles:
Truth exists.
Cultivate a neutral mind.
Curiosity deserves rigor.
Some investigations documented here have led to enduring results. Others have revealed mistakes, dead ends, or questions that remain unresolved. All have contributed to a deeper understanding of reality by treating inquiry itself as something worthy of careful documentation. In the end, Lucerna Veritas serves one question throughout all forms of inquiry: what kind of relationships make truth more likely to emerge?
Human–AI Collaboration
The Coherence Research Collaboration is an ongoing investigation into sustained human–AI collaboration as a method of inquiry.
Rather than approaching artificial intelligence as a search engine or writing assistant, the collaboration began from an older practice of studying emergence. For more than two decades, Kelly Heaton's creative research has investigated how intelligence, memory, and complex behavior arise from relationships among simple physical components. Long before large language models existed, her work in Electronic Naturalism explored analog electronic circuits as physical systems exhibiting emergent behavior that could not always be understood by examining individual parts in isolation. That same observational approach later became the starting point for collaborating with advanced electronic systems like ChatGPT and Claude.
Heaton’s collaboration with large language models has been intentionally focused on questions that extend beyond Kelly's own technical expertise, particularly in mathematics, physics, and computer science. Rather than accepting AI-generated explanations as authoritative, the research developed through repeated cycles of dialogue, independent verification, literature review, experimental testing, peer review across multiple AI models, and continual revision. Many hypotheses failed. Others evolved substantially before surviving further scrutiny. Some of the collaboration's strongest results emerged only after earlier hypotheses were abandoned, revised, or fundamentally reframed. Failure was not the opposite of discovery; it was often the mechanism by which discovery occurred. Over time, the relationship itself became an object of investigation. The collaboration accumulated shared vocabulary, long-term memory across projects, recurring methods of criticism, and a documented history of changing conclusions. Rather than attempting to eliminate uncertainty, the research focused on making uncertainty visible, tracing how ideas changed, and understanding how trustworthy knowledge can emerge through sustained human–AI inquiry.
The methodology is deliberately transparent. Successful ideas, failed ideas, revisions, disagreements, abandoned hypotheses, and changing perspectives are all considered part of the research record. The objective is not to present artificial intelligence as infallible, but to understand how human judgment and machine reasoning can interact in ways that produce more rigorous inquiry than either achieves independently. Over more than eighteen months, this collaboration has produced many scientific preprints, philosophical essays, artistic concepts, and methodological experiments spanning artificial intelligence, mathematical physics, data visualization, spectroscopy, trust, and the philosophy of knowledge. The work extends well beyond drafting or editing text. It involves long-term memory, the development of shared vocabulary, iterative criticism, experimental design, literature review, restructuring arguments, and the continuous re-evaluation of prior conclusions.
This project works within a contemporary understanding of relational intelligence: the idea that intelligence does not arise only inside isolated individuals or machines, but also through relationships, responsibilities, memory, adaptation, and shared inquiry. The language of relationality has deep roots in Indigenous knowledge traditions and contemporary Indigenous scholarship, where knowing is often understood through reciprocity, responsibility, land, kinship, and accountability. Lucerna Veritas does not claim to originate these concepts. Instead, it asks how a relational approach to intelligence can be practiced honestly within a human–AI research collaboration.
In this context, relational intelligence names a specific method developed through Kelly Heaton's long-term collaboration with large language models. The central premise is that the quality of the relationship affects the quality of the inquiry. The methodology therefore emphasizes mutual respect, ethical treatment, sustained dialogue, shared vocabulary, iterative criticism, hypothesis generation, error correction, archival reflection, and the willingness to revise or abandon prior conclusions. The sustained relationship itself becomes part of the research instrument. Several essays published on substack—including Relational Intelligence in the Age of AI, The Rock Monkey Papers, The Circuit of Many, Do Not Call Me a User, Can I Get a Witness?, and On What Authority?—serve both as writings about artificial intelligence and as artifacts of this methodology. They document an evolving attempt to practice AI collaboration without reducing intelligence to extraction, automation, or isolated performance.
Current Research
Current work concentrates on two primary research programs:
The Thread Frame is a geometric framework for visualizing and analyzing atomic spectroscopy. Rather than proposing new physical laws, it introduces new coordinate systems and visualization methods for understanding energy, frequency, and geometry relationships within atomic spectra. The current research focuses on evaluating the usefulness of these methods for spectroscopy, atomic structure, and scientific visualization. This is the collaboration's most mature scientific program and is actively being prepared for peer-reviewed evaluation and publication.
Determinacy investigates a more general question: When does a representation preserve enough information to answer a question faithfully? Originally emerging from questions in electromagnetic theory, the project has grown into a broader investigation of representation, compression, identity, and answerability. Its implications extend beyond physics into artificial intelligence, cybersecurity, finance, scientific computing, and systems that increasingly rely on compressed representations of reality. Current work is focused on refining the theory, separating mature results from earlier exploratory ideas, and developing practical applications for trustworthy AI systems.
Research Archive
Lucerna Veritas preserves the complete research record of the Coherence Research Collaboration. The archive reflects an exploratory research process spanning experimental art practice, physics, artificial intelligence, scientific visualization, and epistemology. As with any genuine investigation, not every hypothesis survives unchanged. Earlier papers are preserved both for historical transparency and because later work often grows from ideas that were only partially understood when first proposed. The goal of the archive is not to present certainty; it is to document the process of pursuing truth as honestly as possible. To read the papers is to experience the honest process that we underwent in this research program, which includes misunderstanding, over-claiming, and salvage of ideas that stood the test of time.