Project Goals

The primary goals of this project are:

• (Phase 1) To Build Critical AI Data Infrastructure: To develop and continuously update a comprehensive, country-level, structured panel dataset on AI, consolidating fragmented data sources to serve as a critical open-science resource for global research. The dataset brings order, logic, and structure to fragmented AI data sources, including structured data from AI index databases, semi-structured data from statistics releases, and unstructured text from country reports. Phase 1 aims to facilitate data scientists, especially social data scientists, to conduct open-source AI research in a time-efficient fashion.

• (Phase 2) To Facilitate Public and Research Discourse: To create a dynamic web application that aggregates and synthesises the global AI data and news across different domains (e.g., society, economy, geopolitics, research, governance), enabling quantitative researchers and the public to gain real-time insights into global AI trends.

Methodology and Phases

Phase 1: Data Compilation and Documentation (The Panel Dataset)

The goal of this phase is to develop Version 1 (v1) (alternatively known as wave 1) of the panel dataset and supporting documentation.

• Data Sources: v1 will compile country-level panel data from three core global databases: the Global Index on Responsible AI, Stanford’s AI Index, and the OECD Artificial Intelligence Policy Observatory. Future versions will include the future waves of panel data released by these databases, and panel data from other sources, including MacroPolo's Global AI Talent Tracker and UNESCO’s Global AI Ethics and Governance Observatory.

• Methodology: I will use a Python script to programmatically retrieve and structure the raw data (structured, semi-structured, and unstructured) from these databases and observatories.

• Output: A complete panel dataset in CSV, STATA, and Excel formats and all necessary supporting documents, will be deposited in Harvard Dataverse, facilitating time-efficient, open-source AI research.

Phase 2: The Interactive Data Hub (The Web App)

The goal of this phase is to build a web app to aggregate data, make it public, and provide advanced analytics.

• Technical Stack: I will use three separate API keys for this intelligence layer:

  • Core AI API (Gemini 2.5 Pro): To analyse and classify all incoming data, perform data synthesis, and generate dynamic, evidence-based reports and summaries.

  • NewsData.io Free News API (Data Aggregation API): To syndicate and retrieve real-time AI-related news and articles from leading media.

  • Mapbox Web Services API (Visualisation API): To create the interactive data dashboard, including dynamic charts and geospatial mapping of country-level AI indices, enabling a user-friendly exploration of global AI trends.

• Methodology: The web app will synthesise both the clean, structured data from Phase 1 (the panel dataset) and other relevant data (structured, semi-structured, or unstructured) from the aforementioned observatories that are not featured in my panel dataset, alongside dynamically analysed unstructured text from real-time news sources. This provides a dynamic, evidence-based view of global AI trends and their impacts.

(Note: Both phases are expected to be continuous in the long term, with periodical updates to the dataset and the application.)

Timeline & How will this funding be used?

I am seeking nine-month funding support for my development, delivery, and launch of the panel dataset (Phase 1) and web app (Phase 2).

• Timeline: The project requires nine months to ensure the necessary manual cleaning and testing is completed.

  • Phase 1: Data Compilation and Documentation (Months 1–5): Python script setup, data collection, manual cleaning, verification, consolidation, and publication.

  • Phase 2: Web App Development (Months 6–9): Prototyping, development, content integration, public beta testing, polishing, and final launch.

Funding Requested: US$32,000

• Stipend US$30,000 (for 9 months): I understand that this request is far lower than the average salary level of a newly qualified postdoc in the UK or the US. However, to save on living expenses, I relocated to Southeast Asia in August 2025. Therefore, US$30,000 would be sufficient to cover my living expenses. Part of this stipend will be used to cover my necessary visa- and travel-related expenses. I aim to start this project by working at 1.0 FTE. However, in the coming months, if I have other research responsibilities to assume, I will adjust to 0.8 FTE. That said, I will ensure the completion of Phase 1 by month 5, and Phase 2 by month 9.

• API costs US$2,000: Despite being university-unaffiliated as an independent researcher, I believe I remain eligible to apply for US$1,000 credits designated for PhD researchers from Google Cloud since my PhD profile is still browsable on the Cambridge department’s website. That said, I would like to request US$1,600 to cover the expenses for using the NewsData.io API Basic Plan (US199.99/m) and request US$400 to cover the expenses for using the Mapbox Web Services API (to cover potential usage overage beyond the free tier).

Who is on your team? What's your track record on similar projects?

I, Jason Hung, am the only person designing, developing, and executing this project. I am a recently graduated PhD in Sociology from the University of Cambridge. I was trained in quantitative social science research for 10 years (PhD, Cambridge; MA, UCL; MSc, LSE, BA, Warwick).

Relevant Track Record (Pre-Doctoral Fellowships, Applied Research and Internships): I have worked independently and collaboratively across diverse domains, demonstrating strong cross-disciplinary capacity:

• Social and gender justice: Institute for Humane Studies (across Thailand, 2023-2024), United Nations (Bangkok, 2019) 

• Social epidemiology/public health: UNU-Institute for Global Health (Kuala Lumpur, 2022), UC Berkeley School of Public Health (Berkeley, 2018) 

• Education and international development: Academia Sinica (Taiwan, 2022-2023), Stanford University (Palo Alto, 2019) 

• Geopolitics: Harvard University Asia Centre (Boston, 2022-2023), Pacific Forum (remotely, 2019-2020) 

• Socio-technology: Asian Development Bank (Manila, 2022)

Similar Projects:

• Competing for AI Talents: Comparative Analysis of AI Adoption in Asia’s Labour Markets (Published, Oxford Intersections, Oxford University Press).

• Bridging AI and Inequalities Research: A Metascience Framework for Methodological Pluralism (Submitted grant application, institutionally supported by the Department of Informatics, KCL).

What are the most likely causes and outcomes if this project fails?

• Data Source Erosion (Technical) Problem: Any global AI index databases or observatories I am using (Phase 1) that change their website structure may break my Python scraping scripts and halt data flow.

• Time-Related (Operational) Problem: If the data cleaning (Phase 1) is far more complex and laborious than anticipated, this may delay the launch of the web app (Phase 2). Therefore, in v1 of compiling and documenting the global AI panel dataset, I am only featuring data from three AI databases/observatories. If v1 of the panel dataset can be delivered on time, I will scale it up by including more AI databases/observatories in the following versions/waves.

(Note: The funding requested will be used for the development, delivery, and launch of Phase 1 and Phase 2 of my project. Once the expected outputs are delivered for both phases, I would have a more complete, convincing profile and body of work in AI-related work to seek funding from EA Long-Term Future Fund, Open Phil, or other similar funders to financially support the continued operation of both Phase 1 and Phase 2 of my project in the long term.)

How much money have you raised in the last 12 months, and from where?

• No.

• I applied for the EA Long-Term Future Fund and Open Phil’s Career Development and Transition Funding earlier this year, but my applications were unsuccessful. Upon speaking with an adviser from 80,000 Hours, I was advised that I should build and enrich my AI-related portfolio of work, which will increase my chance to secure funding from EA Funds, Open Phil, or their equivalents in the long term. 

• I developed this project idea (both Phase 1 and Phase 2) in July 2025 as I realised the data gap that required me to laboriously and manually extract and transport panel data from multiple AI databases/observatories to create my own datasets whenever I conduct AI data research. 

• Therefore, I am developing this project to (1) facilitate open source AI research in a time-efficient manner (both phases), (2) enrich public understanding of global AI trends (Phase 2), and (3) help time- and cost-efficiently inform policymaking (both phases).