Two Ways to Help Us Solve Hard Problems
The Consortium builds peer-reviewed, openly licensed knowledge on humanity’s hardest problems. You can contribute by authoring AI-assisted documents yourself, or by funding the research team that drafts foundational work. Both paths feed the same open library.
Write with AI as your research partner
Pick a problem, open the editor, and draft a document. Connect your own AI API key to accelerate research, synthesis, and critique. Your key stays in your browser — we never store it. Peer reviewers hold your draft to the same rubric regardless of which model (or no model) you used.
- Free to join — GitHub or Google sign-in
- Bring your own key, or draft without AI
- Keep attribution and earn contribution badges
Fund the team’s AI research work
Prefer to help with money rather than time? Donations are the primary way our small research team buys model credits and contractor hours to draft foundational documents on problems the community hasn’t yet covered. Everything the team produces is peer-reviewed and published openly.
- 80% of every dollar goes to AI research & drafting
- One-time, recurring, or problem-earmarked
- Transparent allocation, published budget
From a blank page to a peer-reviewed document
Five steps, most of which you can do in a single sitting. You own your drafts until you submit them. You can stop at any step and come back later — nothing is lost.
Pick a Problem
Browse our curated list of global challenges scored by Importance, Tractability, and Neglectedness (ITN). Every problem has a checklist of documents it needs — from problem definitions and literature reviews to policy briefs and solution roadmaps. Pick the gap that matches your background or curiosity.
Draft with AI (or without)
Open the editor and optionally connect your own API key for Anthropic, OpenAI, Gemini, Mistral, or xAI. Keys stay in your browser and are never stored on our servers. The AI is a research accelerant — you still decide what the document says. You can also draft the old-fashioned way; the markdown editor works fine without AI.
Submit for Peer Review
When your draft is ready, submit it to the editorial queue. Community members and domain experts leave structured feedback — what to fix, what to cite, where the argument breaks. You revise as many times as needed before a document moves to formal review.
Editorial Acceptance
Trained reviewers evaluate submissions against our quality rubric: accuracy of claims, quality of sourcing, clarity for a policy-literate reader, and tractability of proposed actions. Accepted documents are published openly under Creative Commons and added to the problem's progress checklist.
Earn Recognition
Contributions earn badges, climb the leaderboard, and appear on your public profile. When a problem's full document suite is complete, every contributor who helped close it receives a permanent 'Problem Solved' badge. Your work keeps getting cited for as long as the platform exists.
What the AI is actually good for
The AI panel is a research tool, not a ghostwriter. It accelerates the parts of research that are mechanical — gap-finding, synthesis, audience rewrites, structured critique — and leaves the judgement calls to you. Here are four concrete uses contributors rely on most:
Gap-finding
Paste your draft and ask the model to list missing arguments, unaddressed counterpoints, and weak citations. Treat its output as a checklist — you still judge which gaps matter.
Literature synthesis
Feed in a batch of abstracts and ask for a comparative synthesis. The model is a reading accelerant, not a source of truth — verify every factual claim against the primary source.
Audience rewrites
Rewrite the same section for a policymaker, a funder, or a domain researcher without changing the argument. Useful when the same work has to be legible to the people who can act on it.
Structured critique
Ask the AI to steelman an opposing view, red-team your conclusion, or audit a section for unstated assumptions. Use the output as a starting point — never as a verdict.
We never send your API key to our servers at rest. Requests are proxied so we can enforce rate limits and log usage anomalies, but the key lives in your browser’s local storage and you control when to revoke it.
How your donation becomes a published document
Donations don’t vanish into a general fund. They pay for model credits and researcher time against a specific, published budget — and the resulting documents go through the same peer review as every other submission.
You donate
One-time or recurring via Stripe. Every dollar is allocated against a published budget: 80% AI research and drafting, 15% platform and security, 5% operations.
Team buys compute and time
The research fund pays for model credits across Anthropic, OpenAI, Google, Mistral, and xAI, plus contractor time for researchers who specialise in global-priority problems.
Documents go through peer review
Team-authored drafts enter the same editorial queue as community contributions. The same rubric applies to everyone — no special treatment for funded work.
Accepted work is published openly
Final documents are released under Creative Commons Attribution 4.0 — free for anyone to read, cite, or translate. Donors are credited in an aggregate acknowledgement on each document.
Where every dollar goes
Model credits across providers, plus contractor hours for researchers with subject-matter expertise.
Hosting, monitoring, backups, and security audits that keep the site open and trustworthy.
Compliance, accounting, and the minimum legal overhead required of a non-profit.
The 10 Standard Document Types
Every problem in the Consortium requires the same structured set of documents, modelled on IETF RFCs. This ensures completeness and makes it easy for policymakers and funders to find the information they need.
Frequently Asked Questions
Do I need academic credentials to contribute?
No. We value clear thinking, honest sourcing, and intellectual humility over formal titles. Many of our best contributors are autodidacts, practitioners, or students with a strong grasp of a specific problem.
What if I don't want to use AI at all?
That's fine. The editor is a markdown editor first; the AI panel is optional. Peer review is identical either way — reviewers judge the document, not the tools used to produce it.
How much will AI API keys cost me?
It depends on the provider and model, but a complete draft typically uses $0.20–$2 of Claude Sonnet or GPT-4o, and much less with smaller or open models. You pay the provider directly; we take nothing.
How much time does a contribution take?
A solid draft usually takes 2–8 hours depending on scope. A focused section, citation audit, or literature scan can be done in a single sitting. Every contribution counts, no matter the size.
Are documents truly open access?
Yes. Accepted documents are published under Creative Commons Attribution 4.0 International (CC BY 4.0) — free to read, share, translate, and adapt with attribution. No paywalls, no sign-in required to read.
Where does donated money actually go?
80% to AI research and drafting (model credits, contractor research time), 15% to platform infrastructure (hosting, security, monitoring), and 5% to operations (compliance, banking, accounting). A detailed breakdown is on the donate page.
How long until my submission is accepted?
Most drafts receive first-pass reviewer feedback within about two weeks of submission. Acceptance depends on revision cycles — how many changes are requested and how quickly you address them. Quality takes priority over speed.
Can organisations sponsor specific problems?
Yes, within editorial independence. Earmarked donations can be directed to specific problems or document types, but the same peer-review rubric applies and the Consortium retains full editorial control. Reach out via the contact page to discuss.
Pick whichever path fits you today
Both paths feed the same open library. You can switch between them at any time — write this month, donate next, or do both.