About
About this experiment
The idea
Australia's news is full of problems that everyone agrees are problems: tobacco shops burning, programs blowing their budgets, black markets growing. The reports, audits and figures that explain them are public but scattered across hundreds of PDFs. We wanted to see what happens when you hand all that to an AI and ask a simple question: what's going on, why, and how would you fix it?
How a page is made
1. A human picks a topic from the current Australian news.
2. An AI model searches for and reads the published public sources: audits (ANAO, VAGO), Commonwealth and state budget papers, agency quarterly reports, Senate estimates, Productivity Commission, ABS, AIHW, tender data, police and agency media releases, and established news outlets.
3. The AI writes the page in three parts: what's happening (facts, each with a linked source and a date), why it's happening (the incentive or gap, with figures), and how to fix it (ranked, concrete, 3 to 5 items).
4. A human checks every number against its linked source. If a number cannot be sourced, the page says 'no public figure' instead of inventing one.
5. The page goes up with the disclaimer banner, the author line, and a corrections link.
The rules we follow
Public bodies, not people. We critique policies, programs, departments and agencies. We never name or characterise private individuals, and we do not comment on open criminal matters beyond what police and courts have published.
Facts stated, sources linked. Every opinion sits next to the facts it rests on, so you can check them.
No invented numbers. No source, no number.
Fast corrections. 48 hours, noted on the page.
Savings estimates
Each issue page carries a 'What fixing it could save' range in dollars a year, and the home page adds them up. The method is the same every time: start from a government figure (linked), apply two to four plainly stated assumptions, each labelled OUR estimate, and show a low and a high case. They are not government projections or costings. They exist so the size of the prize can be argued about with the assumptions in the open; if you think an assumption is wrong, tell us and we will re-run the numbers.
What this is not
It is not journalism, and it is not advice. It is an experiment in whether an AI can be genuinely useful on hard public problems, published for fun. Read the disclaimer.
Who
Built by an independent hobbyist in Melbourne with an AI model. Not affiliated with any government, party, company or media outlet. Contact: [email protected].