WEBINAR 2 12 Dec 2024 07:00 - 10:00am (GMT-4)
This AI and Data Science Sprint for Economic Statistics plays an important role in the ongoing efforts of the statistical community to integrate advanced data science techniques within the framework of national and international statistical systems. By highlighting successful use cases and discussing the strategic integration of new technologies, the sprint will help in shaping the future of economic statistics production. This Sprint will also be an important step in the further development of the Data Science Leadership Network playbook and seeks to empower NSOs with cutting-edge tools and methodologies for statistical production.
The sprint has four main objectives: (1) Present a series of impactful AI and data science projects in statistical production utilizing reproducible analytical pipelines; (2) Present AI and data science projects for the compilation of economic indicators to inform economic policies; (3) Present the possible use of generative AI technologies for the dissemination and interpretation of statistics; and (4) Address strategic and cross-cutting issues.
Presentation Annotations Agenda
Session 3
Leveraging Open-Source Language Models for Automatic Codification of Employment and Economic Activity in Household Surveys
Alejandro Pimentel, Deputy Director of Research in Data Science, INEGI, Mexico
Summary: This presentation will outline how INEGI is utilizing and investigating open-source language model architectures to automate the codification of responses in household surveys and censuses. By implementing these models, INEGI can efficiently classify employment status and economic activity, significantly reducing manual labor and minimizing errors.
Session 4
IntelliStatcan: Gen-AI chatbot for obtaining statistical information
Milana Karaganis, Director General, Digital Strategic Services
Summary: As part of Statistics Canada’s roadmap for AI adoption, the agency has been experimenting with Gen-AI technologies and methods to address important business problems. This presentation will provide an overview of IntelliStatCan, a Gen-AI powered chatbot to help visitors of Statistics Canada web-site find information within publications available on the web-site. The presentation will discuss the business opportunity, the approach, methods and technology used, and lessons learned to-date.
Automatic Generation of Informative Documents Based on Official Statistics
Amado Esquer, Chief of Research in Data Science, and Elio Villaseñor, Director of Data Science Lab, INEGI, Mexico
Summary: This presentation will discuss how INEGI is evaluating the use generative AI to automatically produce documents based on official statistics. By utilizing this technology, it is possible to generate statistical reports and summaries quickly and consistently, ensuring high-quality outputs that are both accurate and informative for a broad audience.
Building Trust in AI: Statistics Canada's Framework for Responsible and Ethical Governance
Christos Sarakinos, Director of Data Science and Innovation at Statistics Canada
Summary: Statistics Canada’s approach to responsible AI combines structured oversight with ethical standards, aligning AI practices with global frameworks and emphasizing transparency, fairness, and accountability. This presentation highlights the agency’s commitment to building public trust through rigorous peer reviews, collaborative partnerships, and comprehensive governance in AI ethics.