You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
I am seeking USD 24,000 to complete a four-month stage of my ongoing Generative Economics research. The programme asks how human capability and AI autonomy together determine what people can produce. This request supports work already underway: further checking of the economic model, a focused evaluation of the human effort needed to verify AI research outputs, and completion and dissemination of the resulting paper. The minimum viable award is USD 12,000 for a narrower three-month stage.
The central question is how much useful work remains after an AI-generated answer has been checked and corrected. A researcher with limited time may gain little from producing many more drafts if every draft requires extensive checking. I want to make that constraint measurable and connect it to decisions about how much autonomy to give an AI system.
My existing theoretical work and research data experience provide the starting point. The evaluation described here has not yet been conducted. Funding will pay for future work and publication expenses; I am not asking to reimburse past spending or presenting existing manuscripts as new work.
I will develop the verification component of the ongoing economic model and specify which measurements would support or challenge it. The focus is the relationship between the number of candidate outputs, human checking capacity, and reliable output. I will use the results to revise one core paper rather than promise to complete the entire five-paper programme with this grant.
For the empirical evaluation, I will construct 120 tasks from public research-funding records, beginning with NIH RePORTER. Tasks will cover factual extraction, summaries of stated aims, and checking whether sources support a claim. Each task will have a reference answer and source identifiers, with an unresolved category for ambiguity.
I will compare two AI systems under two workflows, with two runs per configuration: 960 outputs in total. One workflow will require evidence for material claims and explicit uncertainty; the other will produce a direct response from the same source packet. At least 120 outputs, balanced across conditions, will receive timed human review. A second paid reviewer will check a subset. I will record correction time, unsupported claims, remaining errors and API costs, with model versions and prompts retained.
At full funding, month one covers the model revision and 30-task pilot, months two and three cover the evaluation, and month four covers analysis, the revised paper and public release. Deliverables are one working paper prepared for journal submission, the evaluation protocol and original code and annotations, and a short practical account of what changes the cost of checking AI work. Source data will be shared only where redistribution is permitted.
At USD 12,000, I will narrow the scope to a 30-task pilot, 240 AI outputs and at least 40 timed reviews, alongside a focused revision of the core paper. That three-month version retains paid researcher time and a publication reserve. The larger award adds the full task set and more extensive model checks.
The USD 24,000 budget comprises USD 12,000 for 160 hours of my research time at USD 75 per hour; USD 3,000 for 120 research-assistant hours at USD 25; USD 1,000 for 20 independent-review hours at USD 50; USD 1,000 for model access; USD 1,000 for data, storage and the reproducible release; and USD 6,000 reserved for submission fees and article processing charges. My commitment averages about ten hours per week over four months.
The USD 12,000 minimum comprises USD 6,000 for 80 hours of my time; USD 1,000 for 40 assistant hours; USD 500 for ten review hours; USD 500 for model access; USD 500 for the data and public release; and USD 3,500 for submission fees and article processing charges. Researcher compensation is a core cost at both levels. These amounts are gross project allowances, not a promise of a particular after-tax income.
Publication amounts are budget reserves, not quotations from a named journal. They cover prospective submission fees and article processing charges only for outputs within this project, subject to the grant agreement. I will check institutional coverage and waivers before paying and will not charge an expense already covered elsewhere. An unused reserve will be returned or reallocated with the funder's approval. Preprints and permitted research materials will be released even if journal review takes longer than the project.
I will report progress after the pilot and at project completion. The proposal requests a research grant, with the payment route and eligible costs agreed before funds are accepted.
I am Hana Kim, a contract research professor in KAIST's School of Business and Technology Management in South Korea. I hold a PhD in Technology Management, Economics and Policy and a master's degree in medicine from Seoul National University. My research combines quantitative analysis with the construction and checking of research datasets. I will lead this work from Korea; any research assistant or independent reviewer will be recruited if funding is secured.
My recent sole-authored research includes a Policy Sciences article on coordination in global AI governance and a Health Policy and Technology article on the quality and interpretation of pharmaceutical authorisation records. These studies combine policy analysis, dataset construction and explicit checks on what the evidence can support. My current Generative Economics work studies the role of human capabilities as AI becomes more autonomous. This project gives that programme a small, auditable empirical component and time to bring a core paper through revision and submission.
Reference answers may be ambiguous, review times may vary substantially across reviewers, or model updates may change results. The pilot will test the scoring protocol before expansion. I will retain ambiguity, document model versions and report reviewer-specific timing. These data describe the defined tasks and reviewers; they do not by themselves estimate productivity for an occupation or establish a causal effect of AI adoption.
The study may find that structured checking costs more than it saves. That result will still be reported. Journal acceptance is outside the project's control, so the committed outputs are a completed working paper, documented methods and accessible research materials, with journal submission funded where appropriate.
I have also submitted an Emergent Ventures request for USD 30,000 for a six-month stage of the same research programme. The requests overlap and are alternative or coordinated routes to support, not additive charges for the same work. If both attract funding, I will disclose the overlap and agree a revised scope and budget before accepting awards. I will record time and expenses by project.
I am not currently receiving a separate research grant for the work proposed here. I participate in an institution-administered research institute project supported by the National Research Foundation of Korea (NRF). That participation is not presented here as a personal grant award. I will check the allocation of research time and expenses so that any costs covered by the institutional project are not also charged to this grant.
Kim, H. (2026). Coordinating fragmented global AI governance: Control, Rights, and Safety as a regime complex. Policy Sciences. Published online 23 September 2026. DOI: 10.1007/s11077-026-09632-w.
Kim, H. (2026). Source-record correlates of limited observed authorisation-holder diversity in marketed archive records with recorded exclusivity expiry. Health Policy and Technology, 101304. In press. DOI: 10.1016/j.hlpt.2026.101304.
Academic profile: Hana Kim