KERALA RISING · FROM EXPORTING PEOPLE TO EXPORTING IDEAS Appendix G: Questions a reviewer will ask The program chapters answer the objections specific to each program, and Chapter 21 answers the objections to the institution. This appendix consolidates the cross-cutting questions a careful reviewer raises on a first reading, in one place, with short answers and pointers to where the full treatment lives. It is meant to be skimmed by someone deciding whether the longer argument is worth their time. On cost and money Is the AI cost really that low? Yes, and it is checkable. A purpose-trained classifier runs at a fraction of a paisa per inference and a frontier vision model at under two rupees; even a million verifications a month lands in the low lakhs of rupees a year. The per-inference prices are cited and reproducible. See Chapter 5 and Exhibits 14–16. Will it become a permanent subsidy? The structure is built to prevent that: a one-time, milestone-gated capitalisation, shared revenue lines that grow with every program, and a path to operating self-sufficiency traced across three scenarios. The state is asked to capitalise once, not to fund forever. See Chapter 16. 339