About

Dan Brown is an adjunct professor in the Department of Geography at the University of Washington. He serves as the director of the School of Environmental and Forest Sciences, where he plays a vital role in guiding the school's academic growth, developing new initiatives, and providing leadership and management of its programs, centers, and research grants. His responsibilities also include allocating revenues to support the school's mission and enhancing its endowment. Dan Brown sits on the Natural Resources Board of Washington State, which oversees the management of state lands. His recent academic experience includes serving as interim dean for the School for Environment and Sustainability at the University of Michigan, where he was also a professor in conservation ecology and environmental informatics. His research interests focus on land use change and its effects on ecosystems and human vulnerability. His work connects computer-based simulation of land-use-change processes with GIS and remote sensing data on historical landscape change and social surveys.

Research topics

• Computer Science
• Political Science
• Geography
• Economics
• Ecology
• Social Science
• Environmental science
• Natural resource economics
• Sociology
• Biology
• Geology
• Cartography
• Business
• Physical geography
• Remote sensing
• Economic growth
• Public economics
• Oceanography
• Management science
• Law
• Environmental resource management

Selected publications

• AI for social good

  Royal Society Open Science · 2025-06-01

  articleOpen accessSenior author

  This article describes the Generative AI Education ( GenAIE) programme: using generative AI (GenAI) to provide personalized education to disadvantaged people, notably probationers and prisoners. For the UK Probation and Prison Service, GenAI (Introducing ChatGPT, 2025, OpenAI; see https://openai.com/blog/chatgpt (accessed January 2025)) is providing education for felons to help stop them reoffending. The UK has over 80 000 prisoners and education is the best deterrent to reoffending, which costs £18bn ($23b) pa (Reoffending Costs, UK Parliament, 2022; see https://questions-statements.parliament.uk/written-questions/detail/2022-0309/137323 (accessed January 2025)). The AI ‘tsunami’ led by GenAI will be hugely disruptive for business and society. However, it also offers pioneering opportunities for social good, notably through personalized education/training for socially excluded and disadvantaged groups (e.g. people on probation, people in prison, refugees, long-term unemployed, long-term sick, low-aspiration young people); thereby transforming their future and addressing major social problems. As a starting point, University College London and MegaNexus are working with educational professionals to produce personalized training content specifically for the Justice sectors, including probation and prisons, described below, which evidences and demonstrates the positive power of GenAI for social benefit. This is part of our AI for Social Good programme. As of 31 November 2024, the GenAIE programme had gained over 53 400 users and accumulated over 596 600 hours of Education, Training and Employment learning. We are now working with local councils to support their social services key workers and their clients. To make our paper self-contained but concise, key technical terms are defined as bullet points.

  PublisherDOI

• Opportunities and challenges to community-level adoption of natural climate solutions in Washington State

  PLOS Climate · 2025-02-28 · 1 citations

  articleOpen accessSenior author

  Natural Climate Solutions (NCS) are climate mitigation approaches that aim to incorporate sustainable practices in forest, agriculture, wetland, and grassland management to increase GHG mitigation from land sectors and have been estimated to be highly effective from global to local scales. As more state and local governments seek to address climate change using a range of available techniques, the potential of NCS has gained increasing attention. As NCS directly involves land management by a range of actors (such as farmers and landowners) operating within resource-dependent communities (such as those dependent on the forest sector), it also has the potential to significantly alter the socioeconomic conditions and opportunities for these communities, necessitating a critical assessment of how NCS implementation interacts with socioeconomic systems. In this work, we focus on the implementation of NCS in Washington State to support its 2050 net-zero goals. Using a novel research approach, we compare recently estimated NCS potentials along multiple pathways with estimates of county-level socioeconomic sensitivities, exposures, and adaptive capacities to NCS-related changes and highlight the potential challenges that exist. These challenges can significantly limit the estimated GHG reduction and ecosystem co-benefits from NCS if they are implemented without due consideration of potential social interactions. We outline policies that can supplement NCS implementation to support just and equitable approaches that contribute to resilient communities and enhance human wellbeing while mitigating GHG emissions from the natural lands of Washington state.

  PublisherOA PDFDOI

• Energy Physics

  Cambridge University Press eBooks · 2025-01-16

  book-chapter1st authorCorresponding
  PublisherDOI

• Corporate Biodiversity Reporting Can Be Scaled With AI and Earth Observation—But Will Miss the Point Without Guidance From Conservation Scientists

  Conservation Letters · 2025-09-01

  articleOpen accessSenior author

  ABSTRACT New biodiversity and ecosystem reporting frameworks require companies to collect data on multifaceted impacts on complex ecological systems over space and time while offering them limited guidance on how to do so. Artificial Intelligence (AI) and Earth Observation (EO) are powerful tools that can help make this reporting efficient and actionable. However, before companies can fulfill their crucially important role in improving the state of nature, they will need guidance from the scientific community to identify meaningful yet scalable metrics for data collection, responsibly apply AI‐enabled EO to reporting workflows, and empower the reporting workforce.

  PublisherOA PDFDOI

• Author response for "AI for social good"

  2025-02-01

  peer-reviewSenior author
  PublisherDOI

• Assessing Language Models' Worldview for Fiction Generation

  arXiv (Cornell University) · 2024-08-15

  preprintOpen accessSenior author

  The use of Large Language Models (LLMs) has become ubiquitous, with abundant applications in computational creativity. One such application is fictional story generation. Fiction is a narrative that occurs in a story world that is slightly different than ours. With LLMs becoming writing partners, we question how suitable they are to generate fiction. This study investigates the ability of LLMs to maintain a state of world essential to generate fiction. Through a series of questions to nine LLMs, we find that only two models exhibit consistent worldview, while the rest are self-conflicting. Subsequent analysis of stories generated by four models revealed a strikingly uniform narrative pattern. This uniformity across models further suggests a lack of `state' necessary for fiction. We highlight the limitations of current LLMs in fiction writing and advocate for future research to test and create story worlds for LLMs to reside in. All code, dataset, and the generated responses can be found in https://github.com/tanny411/llm-reliability-and-consistency-evaluation.

  PublisherOA PDFDOI

• AI for Social Good

  SSRN Electronic Journal · 2024-01-01

  articleOpen accessSenior author
  PublisherDOI

• Watching Popular Musicians Learn by Ear: A Hypothesis-Generating Study of Human-Recording Interactions in YouTube Videos

  arXiv (Cornell University) · 2024-06-06 · 1 citations

  preprintOpen accessSenior author

  Popular musicians often learn music by ear. It is unclear what role technology plays for those with experience at this task. In search of opportunities for the development of novel human-recording interactions, we analyze 18 YouTube videos depicting real-world examples of by-ear learning, and discuss why, during this preliminary phase of research, online videos are appropriate data. From our observations we generate hypotheses that can inform future work. For example, a musician's scope of learning may influence what technological interactions would help them, they could benefit from tools that accommodate their working memory, and transcription does not appear to play a key role in ear learning. Based on these findings, we pose a number of research questions, and discuss their methodological considerations to guide future study.

  PublisherOA PDFDOI

• A Study on Large Language Models' Limitations in Multiple-Choice Question Answering

  arXiv (Cornell University) · 2024-01-15 · 1 citations

  preprintOpen accessSenior author

  The widespread adoption of Large Language Models (LLMs) has become commonplace, particularly with the emergence of open-source models. More importantly, smaller models are well-suited for integration into consumer devices and are frequently employed either as standalone solutions or as subroutines in various AI tasks. Despite their ubiquitous use, there is no systematic analysis of their specific capabilities and limitations. In this study, we tackle one of the most widely used tasks - answering Multiple Choice Question (MCQ). We analyze 26 small open-source models and find that 65% of the models do not understand the task, only 4 models properly select an answer from the given choices, and only 5 of these models are choice order independent. These results are rather alarming given the extensive use of MCQ tests with these models. We recommend exercising caution and testing task understanding before using MCQ to evaluate LLMs in any field whatsoever.

  PublisherOA PDFDOI

• TruthEval: A Dataset to Evaluate LLM Truthfulness and Reliability

  arXiv (Cornell University) · 2024-06-04 · 2 citations

  preprintOpen accessSenior author

  Large Language Model (LLM) evaluation is currently one of the most important areas of research, with existing benchmarks proving to be insufficient and not completely representative of LLMs' various capabilities. We present a curated collection of challenging statements on sensitive topics for LLM benchmarking called TruthEval. These statements were curated by hand and contain known truth values. The categories were chosen to distinguish LLMs' abilities from their stochastic nature. We perform some initial analyses using this dataset and find several instances of LLMs failing in simple tasks showing their inability to understand simple questions.

  PublisherOA PDFDOI

Recent grants

• Project SLUCE: Spatial Land-Use Change and Ecological Effects at the Rural-Urban Interface: Agent-Based Modeling and Evaluation of Alternative Policies and Interventions

  NSF · $1.4M · 2001–2008

• CNH: Collaborative Research: Spatial Land-Use Change and Ecological Effects (SLUCE): Interactions of Exurban Land Management and Carbon Dynamics

  NSF · $1.5M · 2008–2014

Frequent coauthors

• Arun Agrawal

  29 shared

• Derek T. Robinson

  28 shared

• Rick Riolo

  University of Michigan–Ann Arbor

  21 shared

• Joan Iverson Nassauer

  University of Michigan–Ann Arbor

  16 shared

• Kathleen M. Bergen

  University of Michigan–Ann Arbor

  15 shared

• Ana V. Diez Roux

  Drexel University

  15 shared

• Stephen J. Walsh

  University of North Carolina at Chapel Hill

  14 shared

• Jiquan Chen

  Michigan State University

  13 shared

Labs

• Our Advancement TeamPI

Education

• PhD, Geography

  University of North Carolina at Chapel Hill

  1992