You now have the whole regime model, and everything the analysis takes is already in your hands from the previous lesson and readings: read the three states, price each technique's basket against the regime's price table, and ask which channel (usefulness, evidence, direct) its payoff flows through.
All portfolio answers in this module take a three-part shape:
- The allocation: how the effort splits across the technique menu.
- Where the marginal slice of effort goes, and through which channel.
- Which spend binds: the aisle the team would trade any other for.
Module 2 long exercises
Now run the whole machinery yourself. Each scenario hands you the three states, the misalignment world you're probably in, the capability stage you're approaching, and the plan you start under, asking for the decision the module has been building toward: where should the marginal slice of effort go?
Alongside the three states, each regime names the capability stage it is approaching — in the AI Futures Model's milestones (Automated Coder, Superhuman AI Researcher, Superintelligent AI Researcher) — and the initial plan it starts under, in the sense of Plans A, B, C, and D for misalignment risk: roughly, A = a strong international agreement buys a long slowdown, B = the government treats buying lead time as a national-security priority, C = the leading lab spends its few-months lead on safety without government help, D = leadership isn't taking the risk seriously and a small safety team works with what it has.
Price the baskets then defend the result below.