A5058951556- Artificial Intelligence in Law
- Computational and Text Analysis Methods
- Law, AI, and Intellectual Property
- Remote Sensing in Agriculture
- Evolutionary Game Theory and Cooperation
- Ethics and Social Impacts of AI
- Comparative and International Law Studies
- Topic Modeling
- Climate Change Policy and Economics
- Experimental Behavioral Economics Studies
- Land Use and Ecosystem Services
- Climate change impacts on agriculture
- Climate Change Communication and Perception
- Water resources management and optimization
- Sustainability and Climate Change Governance
- Evolutionary Psychology and Human Behavior
- Opinion Dynamics and Social Influence
- Decision-Making and Behavioral Economics
- Judicial and Constitutional Studies
- Complex Systems and Time Series Analysis
- Climate Change, Adaptation, Migration
- Evacuation and Crowd Dynamics
- Complex Systems and Decision Making
- Machine Learning and Data Classification
- Legal Language and Interpretation
Stanford University
2016-2024
New York University
2014-2023
Brooklyn Navy Yard
2016-2022
Vanderbilt University
2014-2018
New York Law School
2017-2018
Harvard University
2017-2018
Mallinckrodt (United States)
2014
The advent of large language models (LLMs) and their adoption by the legal community has given rise to question: what types reasoning can LLMs perform? To enable greater study this question, we present LegalBench: a collaboratively constructed benchmark consisting 162 tasks covering six different reasoning. LegalBench was built through an interdisciplinary process, in which collected designed hand-crafted professionals. Because these subject matter experts took leading role construction,...
Better understanding of Large Language Models' (LLMs) legal analysis abilities can contribute to improving the efficiency services, governing artificial intelligence and leveraging LLMs identify inconsistencies in law. This paper explores LLM capabilities applying tax We choose this area law because it has a structure that allows us set up automated validation pipelines across thousands examples, requires logical reasoning maths skills, enables test manner relevant real-world economic lives...
Drought threatens food and water security around the world, this threat is likely to become more severe under climate change. High-resolution predictive information can help farmers, managers, others manage effects of drought. We have created an open-source tool produce short-term forecasts vegetation health at high spatial resolution, using data that are global in coverage. The automates downloading processing Moderate Resolution Imaging Spectroradiometer (MODIS) sets training...
Better understanding of Large Language Models' (LLMs) legal analysis abilities can contribute to improving the efficiency services, governing artificial intelligence, and leveraging LLMs identify inconsistencies in law. This paper explores LLM capabilities applying tax We choose this area law because it has a structure that allows us set up automated validation pipelines across thousands examples, requires logical reasoning maths skills, enables test manner relevant real-world economic lives...
Access to seasonal climate forecasts can benefit farmers by allowing them make more informed decisions about their farming practices. However, it is unclear whether realize these benefits when crop choices available have different and variable costs returns; multiple countries programs that incentivize production of certain crops while other are subject market fluctuations. We hypothesize the on farmer livelihoods will be moderated combined impact differing economics changing climate....
Out of nearly 70,000 bills introduced in the U.S. Congress from 2001 to 2015, only 2,513 were enacted. We developed a machine learning approach forecasting probability that any bill will become law. Starting with 107th Congress, we trained models on data previous Congresses, predicted all current and repeated until 113th served as test. For prediction scored each sentence language model embeds legislative vocabulary into high-dimensional, semantic-laden vector space. This representation...
We demonstrate a proof-of-concept of large language model conducting corporate lobbying related activities. An autoregressive (OpenAI's text-davinci-003) determines if proposed U.S. Congressional bills are relevant to specific public companies and provides explanations confidence levels. For the deems as relevant, drafts letter sponsor bill in an attempt persuade congressperson make changes legislation. use hundreds novel ground-truth labels relevance company benchmark performance model. It...
Artificial Intelligence (AI) is taking on increasingly autonomous roles, e.g., browsing the web as a research assistant and managing money. But specifying goals restrictions for AI behavior difficult. Similar to how parties legal contract cannot foresee every potential “if-then” contingency of their future relationship, we specify desired all circumstances. Legal standards facilitate robust communication inherently vague underspecified goals. Instructions (in case language models, “prompts”)...
Although there are multiple causes of the water scarcity crisis in American Southwest, it can be used as a model long‐term problem freshwater shortages that climate change will exacerbate. We examine water‐supply for 22 cities extended Southwest United States and develop unique, new measure conservation policies programs. Convergent qualitative quantitative analyses suggest political conflicts play an important role transition regimes toward higher levels demand‐reduction Qualitative...
Artificial Intelligence (AI) capabilities are rapidly advancing, and highly capable AI could cause radically different futures depending on how it is developed deployed. We currently unable to specify human goals societal values in a way that reliably directs behavior. Specifying the desirability (value) of an system taking particular action state world unwieldy beyond very limited set value-action-states. The purpose machine learning train subset states have resulting agent generalize...
The Prisoner's Dilemma has been a subject of extensive research due to its importance in understanding the ever-present tension between individual self-interest and social benefit. A strictly dominant strategy (defection), when played by both players, is mutually harmful. Repetition can give rise cooperation as an equilibrium, but defection well, this ambiguity difficult resolve. numerous behavioral experiments investigating highlight that players often cooperate, level varies significantly...
Urban water supply systems in the United States are increasingly stressed as economic and population growth confront limited resources. Demand management, through conservation improved efficiency, has long been promoted a practical alternative to building Promethean energy‐intensive infrastructure. Some cities making great progress at managing their demand, but study of policies often regionally focused. We present hierarchical Bayesian analysis new measure urban policy, Vanderbilt Water...
Almost all law is expressed in natural language; therefore, language processing (NLP) a key component of understanding and predicting at scale. NLP converts unstructured text into formal representation that computers can understand analyze. The intersection poised for innovation because there are (i.) growing number repositories digitized machine-readable legal data, (ii.) advances methods driven by algorithmic hardware improvements, (iii.) the potential to improve effectiveness services due...
Large Language Models (LLMs) have demonstrated remarkable performance on various quantitative reasoning and knowledge benchmarks. However, many of these benchmarks are losing utility as LLMs get increasingly high scores, despite not yet reaching expert in domains. We introduce ARB, a novel benchmark composed advanced problems multiple fields. ARB presents more challenging test than prior benchmarks, featuring mathematics, physics, biology, chemistry, law. As subset we set math physics which...
We compare policy differences across institutions by embedding representations of the entire legal corpus each institution and vocabulary shared all corpora into a continuous vector space.We apply our method, Gov2Vec, to Supreme Court opinions, Presidential actions, official summaries Congressional bills.The model discerns meaningful between government branches.We also learn for more finegrained word sources: individual Presidents (2-year) Congresses.The similarities learned Congresses over...
We demonstrate a proof-of-concept of large language model conducting corporate lobbying related activities. An autoregressive (OpenAI's text-davinci-003) determines if proposed U.S. Congressional bills are relevant to specific public companies and provides explanations confidence levels. For the deems as relevant, drafts letter sponsor bill in an attempt persuade congressperson make changes legislation. use hundreds novel ground-truth labels relevance company benchmark performance model. It...
Better understanding of Large Language Models' (LLMs) legal analysis abilities can contribute to improving the efficiency services, governing artificial intelligence, and leveraging LLMs identify inconsistencies in law. This paper explores LLM capabilities applying tax We choose this area law because it has a structure that allows us set up automated validation pipelines across thousands examples, requires logical reasoning maths skills, enables test manner relevant real-world economic lives...
The advent of large language models (LLMs) and their adoption by the legal community has given rise to question: what types reasoning can LLMs perform? To enable greater study this question, we present LegalBench: a collaboratively constructed benchmark consisting 162 tasks covering six different reasoning. LegalBench was built through an interdisciplinary process, in which collected designed hand-crafted professionals. Because these subject matter experts took leading role construction,...
Law could recognize nonhuman AI-led corporate entities.
This project applies machine learning techniques to remotely sensed imagery train and validate predictive models of vegetation health in Bangladesh Sri Lanka. For both locations, we downloaded processed eleven years from multiple MODIS datasets which were combined transformed into two-dimensional matrices. We applied a gradient boosted machines model the lagged dataset values forecast future Enhanced Vegetation Index (EVI). The power raw spectral data products compared across time periods...