OntoEdit AI
OntoEdit is an AI system that identifies and analyzes cognitive widgets—the hidden mental frameworks that shape how we think, argue, and make meaning. Beneath all discourse lie interaction metaphors and conceptual structures that remain largely invisible to us, yet determine what we can think and how we think it. These structures both constrain and enable thought, creating particular "problem spaces" while foreclosing others.
Inquiry operates through inherited metaphors that both aid and hinder progress. Every basic metaphor carries metaphysical assumptions that influence downstream thinking. When we unconsciously adopt certain cognitive widgets, we inherit their embedded assumptions. Process-relational philosopher Alfred North Whitehead termed this "misplaced concreteness." OntoEdit supports meta-cognitive inquiry in humans by:
- Identifying outdated and incorrect metaphysical assumptions no longer accepted at the cutting edge of a field, and facilitating the adoption of up-to-date frameworks. This is particularly important at the intersections of knowledge domains, where outdated frameworks often "sneak in."
- Detecting where known metaphysical assumptions conflict or impede progress in conversations or research. Recognizing these assumptions and generating new starting positions can open fresh avenues of progress.
- Enhancing awareness of the nature of cognitive widgets, that they are temporary, limited, relational tools—not reality itself. Beyond increasing self-awareness and cognitive flexibility, recognizing the distinction between mind and reality can significantly reduce existential stress and confusion.
Developing Novel Futures
Accelerating Scientific Progress
For the Scientific Community
Scientific innovation often stalls because of outdated assumptions embedded in how we frame problems. From AI ethics to systems biology, climate modeling to consciousness research, the limiting factor is bad metaphysics.
This tool is a breakthrough infrastructure for metacognitive science.
It offers the first scalable way to analyze and refine the cognitive architectures underlying research, journals, and policy writing. By surfacing hidden metaphors, logical gaps, and epistemic misalignments, it helps scientists and institutions:
- Make better decisions grounded in coherent worldviews.
- Detect and overcome blind spots across disciplinary boundaries.
- Train the next generation of researchers in first-principles reasoning.
- Improve the integrity and originality of scientific literature.
Open New Scientific Territory
For Scientists
You and your colleagues have been beating their head against a wall for years, maybe decades, with fewer and fewer breakthroughs than your predecessors even though your applying more and more resources. The problem is that metaphors and assumptions that worked before are hitting the limit of their usefulness.
This tool helps you discover new ways forward.
Upload your paper and receive structured feedback on:
- Implicit assumptions driving your framing and argument.
- Outdated or contradictory metaphors (e.g. mind as computation, body as container, emergence as a black box).
- Performative contradictions between your methods and conclusions.
- Alternative frameworks with more explanatory power.
This tool will be especially powerful at the edges of disciplines, including AI, consciousness, ecology, biology, and cognitive science, where new science depends on new thought.
A New Standard for Scientific Rigor
For Science Publishers
Science publishers, including professional journals, popular magazines, and social media platforms, are gatekeepers of scientific integrity. Yet, even articles with strong methods, evidence, and argument will still carry framing errors that weaken their relevance and insight. These aren’t surface errors; they shape the problem space a paper is working in.
This tool elevates editorial review by surfacing what’s usually invisible.
It looks for deep cognitive structures—metaphors, assumptions, and reasoning frames—flagging where concepts are incoherent, misaligned, or could be fruitfully replaced. Publishers and platforms using this tool:
- Raise epistemic standards by filtering not just errors but bad thinking.
- Reduce reviewer load by pre-flagging conceptual weaknesses.
- Improve clarity and interdisciplinarity, especially across philosophy of mind, AI, biology, and complex systems.
- Position themselves as leaders in publishing the next generation of rigorous, original science.
Initial OntoEdit Applications
We will begin by focusing on scientific research and publishers, particularly on specific problems in specific scientific fields where we see the most popular malware encroaching on progress. Scientific texts are especially concept-dense, with well-justified arguments and references to supporting research, which will enable the training of OntoEdit to analyze common assumptions across researchers and their impact on the field overall.
The OntoEdit prototype will allow users to upload a research text or set of texts, excavate the underlying cognitive widgets, and enable users to consciously examine the pros and cons of holding to them. The tool analyzes these materials to reveal the underlying mental models shaping the user's perspective. By making implicit structures explicit, the system helps users discover new growth opportunities and alternative ways of understanding their subject matter.
Because The Divinity School is already strongest at the intersection of cognitive science, the philosophy of mind, and artificial intelligence, and it’s relatively novel territory, we will have a focus here at the beginning. Even with AI being one of the most significant areas of scientific research today, a great deal of unanswered technical, philosophical, ethical, and policy questions remain about how to develop the field further.
A few common Cognitive Widgets with significant downstream impacts, that OntoEdit will be trained in are:
- Linear time → Time as experiential relationships, not distances or any other spatialized metaphor.
Influenced by process thinkers like A.N. Whitehead and Charles Hartshorne. - Additive change → Biological and cultural evolution and scientific development occur mostly through mutations and extinction events, not developmental processes.
Influenced by Stephen Jay Gould, Thomas Kuhn, Lance Gunderson and C.S. Holling, Jean Gebser and Merlin Donald - Newtonian causality → Complex potential is latent in the constantly active flux of changing background relationships.
Influenced by Whitehead, Gilbert Simondon, William Connolly, and Christopher Alexander. - Intelligence as hierarchical complexity → The kind of intelligence that matters cuts through complexity and clarifies things. It has more to do with character and virtues of the heart.
Inspired by and partially a critique of adult development theory, largely advanced by Bonnitta Roy - Emergence → Knowable continuous background processes manifest as discrete forms when certain thresholds are crossed (as in cymatics). Emergence is too often used as a scientific black box for novelty as if it comes from nowhere by magic.
This argument from Bonnitta Roy is especially relevant to advancing the field of complex systems science. - Ultimate symmetries → We often assume that foundational opposites are equally important, but in reality, ultimate categories are asymmetrical. For example, in a world with “many,” you can understand the “one,” but in a world with just “one,” you can never derive the “many.”
Based on Hartshorne’s and C.S. Pierce’s metaphysics of asymmetrical dependence. - Evolutionary anthropocentrism → The next highest form does not descend from the previous highest form, but from a previous relatively simple form. Humans are not the “most evolved creatures;” all organisms are equally evolved, and any one of them could dominate the future.
Stephen Jay Gould was a strong advocate of this insight, but it’s a basic tenet of evolutionary biology. - Stratified reality → Understanding events depends on their relational context and historical causes. Going “deeper” into lower levels of structure (e.g. physics) doesn’t get to the bottom of things.
Challenges ontological reductionism in favor of ecological embedding (Bonnitta Roy, Simondon). - Whole-part confusion → Systems don’t just nest parts into wholes. Parts and wholes, like seeds and trees, precede and give rise to each other in mutually generative processes.
Reflects the design logic of Christopher Alexander and Michael Levin’s work on morphogenesis.
Future iterations of OntoEdit will expand to additional application areas, including policy design, philosophical debate, education, and everyday conversation.
Project Significance
Scientific discovery and its application in technology development has been the single greatest driver of change over the last few hundred years. Now, as transformer architectures (those underpinning large language model chatbots like ChatGPT) continue to prove powerful across multiple fields, national governments, private foundations, and venture capital are all recognizing the value of AI for accelerating scientific discovery and automation.
“AI… acts as a catalyst for scientific breakthroughs and a key instrument in the scientific process. This heralds a new era of accelerated results; it pushes scientific frontiers and produces outcomes beyond the reach of current tools. This acceleration can help us tackle pressing societal challenges like climate change, health, and the green and digital transitions, while keeping Europe at the cutting edge of scientific progress.” - European Commission for Research and Innovation
“The use of Artificial Intelligence (AI) in scientific research is a top priority at the Department of Energy.” - U.S. DOE
The Sloan Foundation speaks not just for its own investment in the space, but also for the philanthropic community as a whole when it says: “It seems inevitable that over the coming years public and private R&D funders will make significant investments both to diffuse and adopt AI technologies, and to solve technical challenges, in the direction of a more heavily AI-mediated research. ” Google.org pledged $20 million to support academic and nonprofit organizations using AI for scientific discovery, and the Astera Institute has dedicated $2.5 billion to use AI to “rebuild the machinery of science for our abundant future.”
In this growing field of experimentation, OntoEdit belongs to a category of tools and products built to help researchers reflect on their hypotheses and suggest alternative explanations or novel directions. Examples include Consensus, Elicit, Leap Labs, Sakana, and Google’s AI Co-Scientist. Across the best documented AI for science tools, Elicit reports over 400,000 monthly users and has shown, in independent studies, to drastically cut literature review times while improving screening precision by more than fivefold. Consensus similarly serves hundreds of thousands of users by delivering fast, citation-backed syntheses from peer-reviewed research. Leap Labs and Google’s AI Co-Scientist have demonstrated domain-level discoveries in meteorology, immunology, and microbiology that might have taken human teams months or years to surface.
Most AI for science tools, including those named, aim to accelerate the production of science: automating experiments, mining literature for evidence, predicting molecular interactions, or generating new models. This is valuable—but it largely operates within the existing conceptual boundaries of a field, while foundational conceptual innovation remains underfunded. Supporting it can yield multiplicative returns across disciplines. OntoEdit is unique in its focus on the metaphysical assumptions scientists bring to the table and on rigorous metaphysical reflection and discovery. OntoEdit can change the hypothesis itself, revealing alternative frames that unlock insights that support new paradigmatic research.
Cognitive widgets are the actual machinery through which reality gets constructed in human experience and the foundations of how we build our worlds. When spiritual traditions clash, when scientific paradigms become incommensurable, when political discourse breaks down, it's typically because different cognitive widgets are creating fundamentally different worlds of meaning that can't interface with each other. OntoEdit AI differentiates models and reality so that people develop more sensitivity and insight into their own and others’ intentions and conceptual assumptions and avoid the suffering comes from mistaking our mental models for reality itself.
OntoEdit is being built in partnership with Bonnitta Roy. The particular metaphysical malware we’ve identified are derived from Roy’s decades of scholarship, teaching, and personal mentoring all of which center on the value of metaphysical flexibility and the critical capacity to distinguish mental models, perception, and reality. OntoEdit is a technology working to scale these capacities.