You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$2,500 left to reach the funding goal.
I am Ammara Aftab, founder and Chairperson of Pro AI Global in Karachi, Pakistan. I am seeking USD 2,500 for a six week exploratory project examining evidence quality and meaningful human review in one institutional AI document analysis workflow. The project would produce a public research brief, synthetic examples and a practical review checklist. It is a small discovery project, not a deployed system, a validated safety intervention or a claim of national impact.
What are this project's goals? How will you achieve them?
The question is whether a simple review process can help institutional practitioners identify unsupported claims, unclear source provenance and situations requiring escalation when using AI generated document summaries. The first task is to review existing resources and determine whether a new checklist would add anything useful. I would narrow the project to one workflow before developing materials.
Week one would review relevant literature and existing guidance. Weeks two and three would seek up to eight consented adult practitioner or expert interviews about review needs. Weeks four and five would develop synthetic examples and a prototype checklist, seeking independent expert criticism where available. Week six would produce a public brief describing the evidence, limitations and whether further work appears justified. These are proposed activities, not completed research.
No institutional partner, interview access, confidential dataset or deployment permission is secured. Recruitment and any pilot would require appropriate consent and agreement. I would not collect confidential institutional documents or use the work to make consequential decisions about patients, learners, employees or other individuals. Public outputs would exclude identifiable interview material unless explicitly authorised. If access is insufficient, the output would explain that limitation rather than invent findings.
How will you know it worked?
Success would mean a clearly scoped question, a documented review of existing resources, transparent synthetic examples, a checklist that has received critical feedback where feasible, and an accessible brief distinguishing findings from assumptions. I would report what was actually completed, whether practitioners found the materials understandable, and whether existing guidance already meets the need. Interview counts alone would not establish improved safety. The project could reasonably conclude that a new intervention is unnecessary or that further research should take a different direction.
The possible connection to longer term AI risk is through institutions retaining the capacity to question and oversee increasingly capable systems. That connection is indirect and uncertain. I do not claim that this study would measure or demonstrate a reduction in catastrophic risk, nor do I claim technical alignment expertise.
How will this funding be used?
The proposed budget is USD 2,500: 1,500 for research and writing time; 300 for interview coordination, transcription and accessibility; 300 for independent expert review if engaged; 200 for tools, connectivity and research access; and 200 for editing and accessible publication. These are estimates, not incurred costs or signed contracts. Funding would reserve research time and support criticism beyond my own perspective. Without funding, I would explore a narrower literature review alongside existing work.
The work would begin after funding, lawful payment arrangements and any necessary ethics or institutional requirements are agreed. The six week schedule is provisional. I would agree an appropriate scope and reporting arrangements before accepting support. The intended public outputs would be available without a paid training requirement; this request is not for marketing or routine company operations.
Who is on your team? What's your track record on similar projects?
I would lead the proposed research. No additional researcher or external expert is currently committed. I hold an MA in Applied Statistics from the University of Karachi and worked as Principal AI Research and Strategist at MEGNUS from January 2020 to December 2023. I founded Pro AI Global in January 2023 and currently lead responsible AI education, institutional readiness and executive education work.
Supplied Government of Sindh acknowledgement letters document AI training in 2025 and 2026, including 101 newly inducted officers and a separate senior leadership audience in 2026. Training through ORIC LUMHS engaged academic and medical professionals. I also serve as AI and Innovation Chair in Rotary District 3271, with reappointment documented for 2026/27. These are practitioner and communication strengths, not evidence that I have already completed a comparable empirical research project.
Professional profile: https://www.linkedin.com/in/ammaraaftab/
Organisation: https://proaiglobal.com/
What are the most likely causes and outcomes if this project fails?
The main risks are limited interview access, duplicating existing guidance, an overly broad initial question, insufficient independent review and weak relevance to the most important AI governance problems. I would mitigate these by reviewing existing work first, narrowing the scope, using synthetic material and seeking criticism before recommending further investment. If recruitment or added value is inadequate, the brief would report that openly. No favourable outcome, partnership, publication acceptance or downstream adoption is promised.
How much money have you raised in the last 12 months, and from where?
No award has been received for this project. Pro AI Global has received no outside investment, as confirmed by the applicant representative. Existing commercial income is separate from grant funding and is not described here as research funding.
The same USD 2,500 exploratory proposal is pending with BlueDot Rapid Grants, the Transformative AI Fund, GTR AI Safety Fund and the Foresight Human Empowerment research grant programme. This Manifund request is an alternative source for the same budget, not an additional USD 2,500 to spend on the same costs. Any award or material status change would be disclosed promptly, and duplicate funding would not be accepted. Related fellowship and research applications are also pending; any overlap in time, budget or outputs would be disclosed and resolved before acceptance. No fellowship selection or grant award is claimed.