OCPSG Research Project (RE 4) · Now Recruiting Research Engineers!

Measuring Fragmentation: Developing an AI-Native Method for Tracking the Foreign Investment of Technology Companies

Hello! Geopolitics is reshaping global tech investment. We’re building an AI pipeline to map it. If you’re an entrepreneurial researcher, we need your help. You’ll build it from scratch, learn how to work with AI models, co-author the paper, and help maintain a public dashboard. Read more and apply below.

Project Overview

Abstract

Geopolitics has influenced how technology is built and where it’s traded for centuries. Measuring Fragmentation, an approved Oxford Computational Political Science Group (OCPSG) research project, seeks to build a novel computational method that competes in accuracy with manual ones to answer the following: how are technology companies shifting foreign direct investment abroad? First, AI is a research tool: the research project is intended to benchmark the ability of LLMs to create industry-grade fact sets. Second, AI is the subject of assessment: the method will be piloted on a selected AI company deemed consequential, given its geopolitical influence. The output will be full original research submitted to a peer-reviewed journal and a public website that the interested public, like scholars, investors, and journalists, can rely on.

Background

Geopolitics guides who global technology companies can hire, what they research, which products end up where—and the future: where they’ll invest next. Finance guides where the world’s engineers work, data centres get built, chips flow, and satellites orbit—all impacting humanity. We live in an era where corporations, specifically technology companies, rival the power of governments to shape geopolitics. Flashpoints around the world—over the Pacific, in the Middle East, and in Eastern Europe—threaten to fragment the future flows of today’s global high-technology economy. Some even warn a “Tech Cold War” lies on the horizon.

Proposed Method and Results

OCPSG identifies a novel opportunity to help add quantified scholarship to largely qualitative, case-based analysis that suggests technology companies are adjusting their future foreign direct investments in response to geopolitics. Recent AI/ML advances, coupled with better open data availability than ever, offer a new opening.

The project seeks to refine an existing methodology within a forthcoming report, Bay Area Abroad: Innovation through Fragmentation, to be published with the Bay Area Council Economic Institute. The report narrows over 2,000 technology companies to 47 consequential Bay Area firms. Its conclusion suggests firms are hedging their future investments, but its methods were manual and relied upon paid databases. fDi Intelligence Global Markets Editor Alex Irwin-Hunt’s one-off LLM analysis of “NVIDIA’s empire” comes closest, but drew no inferences about geopolitics.

An LLM reads open-source data, such as company newsrooms and SEC filings, extracts each project with its source quote, and a person checks every record. Fusing datasets can provide unique insights. Capital IQ tracks deals and acquisitions and fDi Markets tracks greenfield investment announcements. The team will select a pilot company before scaling the method. The expected result is a benchmark: how closely an LLM-built record of the FDI of a sample of technology companies matches manually verified datasets. The aim is to empower the analyst with more data that’s more accurate, not to give them conclusions.

Expected Outputs

There are two expected outputs. The core methodology will be submitted to a peer-reviewed journal; all team members will be named co-authors. Alongside it, the project will develop a public website: a dashboard to allow others to visually track technology flows.

Join the Team

Research Engineer | 3–5 Researchers Total | Remote or Oxford, UK

Above all, you are curious, entrepreneurial, and comfortable failing with others. You should be fascinated by how geopolitics and the global high-technology economy influence each other.

Who You Are

Do not be intimidated by our request for “engineers”. OCPSG welcomes applicants from all disciplines beyond political or computer science. Everyone will contribute to the entire project. You’ll be expected to help each other learn how to build the pipeline that sources open information into tested, verified project records that can be benchmarked against preexisting datasets. The team will also build the website that puts those records in view. This is a great way to get technical.

What You’ll Do

• Collect and Extract: build and run the LLM pipeline that reads company newsrooms and SEC filings and pulls each project with its source quote.
• Verify and Code: verify sources, code project types, and other data hygiene practices.
• Benchmark: measure results against established, trusted datasets, such as fDi Markets.
• Deliver: maintain the open dataset, co-author the paper, and build the website.
• Make Mistakes: everyone will make errors. We’ll figure out why. That is learning.

Key Dates

• Applications close: Sunday, 22 November, 11:59pm (UK time)
• Final decisions due: Wednesday, 2 December
• Program start date: Monday, 7 December

Commitment & Logistics

• The expected commitment is about 6–10 hours per week into the late spring of 2027.
• Minimum one synchronous meeting weekly.
• Remote with in-person options in Oxford, UK.
• This project begins unfunded. We believe that after the team and scope emerge, what institutional support the project requires will naturally follow.
• Opportunities for in-person travel and presentation have arisen in the past.