{"680350":{"#nid":"680350","#data":{"type":"news","title":"AI in Action: One Student\u2019s Journey to Smarter Sustainability Policy","body":[{"value":"\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cp\u003E\u003Cem\u003EWhen Ashley Cotsman arrived as a freshman at Georgia Tech, she didn\u2019t know how to code. Now, the fourth-year Public Policy student is leading a research project on AI and decarbonization technologies.\u003C\/em\u003E\u003C\/p\u003E\u003Cp\u003EWhen Cotsman joined the \u003Ca href=\u0022https:\/\/datasciencepolicy.gatech.edu\/\u0022 rel=\u0022noreferrer noopener\u0022 target=\u0022_blank\u0022 title=\u0022(opens in a new window)\u0022\u003EData Science and Policy Lab\u003C\/a\u003E as a first-year student, \u201cI had zero skills or knowledge in big data, coding, anything like that,\u201d she said.\u003C\/p\u003E\u003Cp\u003EBut she was enthusiastic about the work. And the lab, led by Associate Professor Omar Asensio in the \u003Ca href=\u0022https:\/\/spp.gatech.edu\/\u0022 rel=\u0022noreferrer noopener\u0022 target=\u0022_blank\u0022 title=\u0022(opens in a new window)\u0022\u003ESchool of Public Policy,\u003C\/a\u003E included Ph.D., master\u2019s, and undergraduate students from a variety of degree programs who taught Cotsman how to code on the fly.\u003C\/p\u003E\u003Cp\u003EShe learned how to run simple scripts and web scrapes and assisted with statistical analyses, policy research, writing, and editing. At 19, Cotsman was \u003Ca href=\u0022https:\/\/www.sciencedirect.com\/science\/article\/pii\/S2772424723000069\u0022 rel=\u0022noreferrer noopener\u0022 target=\u0022_blank\u0022 title=\u0022(opens in a new window)\u0022\u003Epublished\u003C\/a\u003E for the first time. Now, she\u2019s gone from mentee to mentor and is leading one of the research projects in the lab.\u003C\/p\u003E\u003Cp\u003E\u201cI feel like I was just this little freshman who had no clue what I was doing, and I blinked, and now I\u2019m conceptualizing a project and coming up with the research design and writing \u2014 it\u2019s a very surreal moment,\u201d she said.\u0026nbsp;\u003Cbr\u003E\u0026nbsp;\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cimg src=\u0022https:\/\/iac.gatech.edu\/sites\/default\/files\/2025-02\/Cotsman2_0.jpg\u0022 alt=\u0022Ashley takes a selfie with a friend in front of a poster presentation at a conference.\u0022 width=\u0022570\u0022 height=\u0022430\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cp\u003E\u003Cem\u003ECotsman, right, presenting a research poster on electric vehicle charging infrastructure, another project she worked on with Asensio and the Data Science and Policy Lab.\u003C\/em\u003E\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u0026nbsp;\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Ch2\u003E\u003Cstrong\u003EWhat\u2019s the project about?\u003C\/strong\u003E\u003C\/h2\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cp\u003ECotsman\u2019s project. \u003Ca href=\u0022https:\/\/appam.confex.com\/appam\/2024\/meetingapp.cgi\/Paper\/53485\u0022 rel=\u0022noreferrer noopener\u0022 target=\u0022_blank\u0022 title=\u0022(opens in a new window)\u0022\u003E\u201cScaling Sustainability Evaluations Through Generative Artificial Intelligence\u003C\/a\u003E.\u201d uses the large language model GPT-4 to analyze the sea of sustainability reports organizations in every sector publish each year.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EThe authors, including Celina Scott-Buechler at Stanford University, Lucrezia Nava at University of Exeter, David Reiner at University of Cambridge Judge Business School and Asensio, aim to understand how favorability toward decarbonization technologies vary by industry and over time.\u003C\/p\u003E\u003Cp\u003E\u201cThere are thousands of reports, and they are often long and filled with technical jargon,\u201d Cotsman said. \u201cFrom a policymaker\u2019s standpoint, it\u2019s difficult to get through. So, we are trying to create a scalable, efficient, and accurate way to quickly read all these reports and get the information.\u201d\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u0026nbsp;\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Ch2\u003E\u003Cstrong\u003EHow is it done?\u003C\/strong\u003E\u003C\/h2\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cp\u003EThe team trained a GPT-4 model to search, analyze, and see trends across 95,000 mentions of specific technologies over 25 years of sustainability reports. What would take someone 80 working days to read and evaluate took the model about eight hours, Cotsman said. And notably, GPT-4 did not require extensive task-specific training data and uniformly applied the same rules to all the data it analyzed, she added.\u003C\/p\u003E\u003Cp\u003ESo, rather than fine-tuning with thousands of human-labeled examples, \u201cit\u2019s more like prompt engineering,\u201d Cotsman said. \u201cOur research demonstrates what logic and safeguards to include in a prompt and the best way to create prompts to get these results.\u201d\u003C\/p\u003E\u003Cp\u003EThe team used \u003Cstrong\u003Echain-of-thought prompting,\u003C\/strong\u003E which guides generative AI systems through each step of its reasoning process with context reasoning, counterexamples, and exceptions, rather than just asking for the answer. They combined this with \u003Cstrong\u003Efew-shot learning \u003C\/strong\u003Efor misidentified cases, which provides increasingly refined examples for additional guidance, a process the AI community calls \u201calignment.\u201d\u003C\/p\u003E\u003Cp\u003EThe final prompt included definitions of favorable, neutral, and opposing communications, an example of how each might appear in the text, and an example of how to classify nuanced wording, values, or human principles as well.\u003C\/p\u003E\u003Cp\u003EIt achieved a .86 F1 score, which essentially measures how well the model gets things right on a scale from zero to one. The score is \u201cvery high\u201d for a project with essentially zero training data and a specialized dataset, Cotsman said. In contrast, her first project with the group used a large language model called BERT and required 9,000 lines of expert-labeled training data to achieve a similar F1 score.\u003C\/p\u003E\u003Cp\u003E\u201cIt\u2019s wild to me that just two years ago, we spent months and months training these models,\u201d Cotsman said. \u201cWe had to annotate all this data and secure dedicated compute nodes or GPUs. It was painstaking. It was expensive. It took so long. And now, two years later, here I am. Just one person with zero training data, able to use these tools in such a scalable, efficient, and accurate way.\u201d\u0026nbsp;\u0026nbsp;\u003Cbr\u003E\u0026nbsp;\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cimg src=\u0022https:\/\/iac.gatech.edu\/sites\/default\/files\/2025-02\/Cotsman_0.jpg\u0022 alt=\u0022Cotsman posing in front of the US Capitol building in Washington, DC.\u0022 width=\u0022570\u0022 height=\u0022430\u0022\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cp\u003E\u003Cem\u003EThrough the Federal Jackets Fellowship program, Cotsman was able to spend the Fall 2024 semester as a legislative intern in Washington, D.C.\u003C\/em\u003E\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u0026nbsp;\u003C\/div\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Ch2\u003E\u003Cstrong\u003EWhy does it matter?\u003C\/strong\u003E\u003C\/h2\u003E\u003C\/div\u003E\u003Cdiv\u003E\u003Cp\u003EWhile Cotsman\u2019s colleagues focus on the results of the project, she is more interested in the methodology. The prompts can be used for preference learning on any type of \u201cunstructured data,\u201d such as video or social media posts, especially those examining technology adoption for environmental issues. Asensio and the Data Science and Policy team use the technique in many of their recent projects.\u003C\/p\u003E\u003Cp\u003E\u201cWe can very quickly use GPT-4 to read through these things and pull out insights that are difficult to do with traditional coding,\u201d Cotsman said. \u201cObviously, the results will be interesting on the electrification and carbon side. But what I\u2019ve found so interesting is how we can use these emerging technologies as tools for better policymaking.\u201d\u003C\/p\u003E\u003Cp\u003EWhile concerns over the speed of development of AI is justifiable, she said, Cotsman\u2019s research experience at Georgia Tech has given her an optimistic view of the new technology.\u003C\/p\u003E\u003Cp\u003E\u201cI\u2019ve seen very quickly how, when used for good, these things will transform our world for the better. From the policy standpoint, we\u2019re going to need a lot of regulation. But from the standpoint of academia and research, if we embrace these things and use them for good, I think the opportunities are endless for what we can do.\u201d\u003C\/p\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003EWhen Ashley Cotsman arrived as a freshman at Georgia Tech, she didn\u2019t know how to code. Now, the fourth-year Public Policy student is leading a research project on AI and decarbonization technologies.\u003C\/p\u003E","format":"limited_html"}],"field_summary_sentence":[{"value":"When Ashley Cotsman arrived as a freshman at Georgia Tech, she didn\u2019t know how to code. Now, the fourth-year Public Policy student is leading a research project on AI and decarbonization technologies."}],"uid":"35766","created_gmt":"2025-02-10 19:35:40","changed_gmt":"2025-02-12 19:39:48","author":"dminardi3","boilerplate_text":"","field_publication":"","field_article_url":"","location":"Atlanta, GA","dateline":{"date":"2025-02-10T00:00:00-05:00","iso_date":"2025-02-10T00:00:00-05:00","tz":"America\/New_York"},"extras":[],"hg_media":{"676251":{"id":"676251","type":"image","title":"pics (3).jpg","body":null,"created":"1739217209","gmt_created":"2025-02-10 19:53:29","changed":"1739217209","gmt_changed":"2025-02-10 19:53:29","alt":"Ashley at the US Capitol Building. ","file":{"fid":"259996","name":"pics (3).jpg","image_path":"\/sites\/default\/files\/2025\/02\/10\/pics%20%283%29.jpg","image_full_path":"http:\/\/hg.gatech.edu\/\/sites\/default\/files\/2025\/02\/10\/pics%20%283%29.jpg","mime":"image\/jpeg","size":916905,"path_740":"http:\/\/hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/2025\/02\/10\/pics%20%283%29.jpg?itok=uN_SDgrY"}}},"media_ids":["676251"],"groups":[{"id":"1281","name":"Ivan Allen College of Liberal Arts"},{"id":"1188","name":"Research Horizons"},{"id":"1289","name":"School of Public Policy"}],"categories":[{"id":"151","name":"Policy, Social Sciences, and Liberal Arts"},{"id":"8862","name":"Student Research"}],"keywords":[{"id":"192863","name":"go-ai"},{"id":"187915","name":"go-researchnews"}],"core_research_areas":[{"id":"193655","name":"Artificial Intelligence at Georgia Tech"}],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[{"value":"\u003Cp\u003E\u003Ca href=\u0022mailto:dminardi3@gatech.edu\u0022\u003EDi Minardi\u003C\/a\u003E\u003C\/p\u003E\u003Cp\u003EIvan Allen College of Liberal Arts\u003C\/p\u003E","format":"limited_html"}],"email":["dminardi3@gatech.edu"],"slides":[],"orientation":[],"userdata":""}}}