Let me synthesize the notes into one Markdown note with a single durable claim.

The core question: Do universities grow or shrink due to AI? Regional facet: SEC growth vs northeast/rust belt decline.

Notes 1&2: bifurcation by 2045 — top ~100 grow, middle hollows into teaching-only, bottom third merges/closes. Demography is load-bearing wall; AI is accelerant/sorter. Regional decline tracks demographic map. HS grads peak ~3.9M near 2025–26; post-2008 birth dearth hits 2029–2037; Northeast/Midwest/Plains lose 10–20% college-age; Texas/Florida/Tennessee/Georgia/Carolinas grow.

Note 3: birth figures 2007 ~4.32M → 2023 ~3.59M (~17% decline). Two-front squeeze: fewer 18-year-olds AND degree signal weakened (skills directly verifiable). Regional map: hardest hit Northeast/Midwest small privates + open-enrollment regional publics; most resilient SEC/Sun Belt flagships + super-selectives (demand prestige-elastic, not cohort-elastic). Adaptations: admit-standard relaxation, international recruitment contingent on visa policy, credential repackaging. Watch: endowment repurposing tell.

Note 4: session planning only — merge Notes 1–3.

Note 5: Program mix (licensed vs unlicensed share of credit hours) is dominant predictor; region governs whether winning mix is reachable. Licensed expansion (nursing/engineering) takes 7–10 years — longer than enrollment runway of declining-region schools, so decline locks in. SEC/Sunbelt publics grow seats 10–25% through 2040. Northeast small privates without licensed core shrink 30–50% by 2040. Surviving Northeast exceptions: licensed-core with hospital-system/land-grant ties. Formulation: "program mix picks the winners, regional demography picks which losers can still switch."

Note 6: Refinement of Note 5 — region modulates feasible program mix, not just growth. Two mechanisms: demographic base + pivot lag. Falsification test: if Northeast pivot attempts succeed at rates comparable to Sunbelt licensed launches, pivot-lag mechanism wrong.

The most evolved, specific claim from the last steps (Note 5/6) is the program-mix × region interaction. But the durable synthesized claim across the session should integrate the bifurcation claim with the newer, more precise mechanism: program mix picks winners, demography picks which losers can still switch.

I think the best note combines:
- Core: bifurcation by 2045, not uniform shrink.
- Mechanism evolution: demography is the load-bearing wall; AI is the accelerant; the newest refinement is that program mix (licensed vs unlicensed) picks winners and regional demography picks which losers can still switch (pivot lag 7–10 years vs enrollment runway).
- Numbers with attribution: births 2007 ~4.32M → 2023 ~3.59M (~17%) — source: CDC/NCHS birth data (that's the actual source of those figures; Note 3 presented them). HS grads peak ~3.9M around 2025 — that's WICHE projections. I can name these sources since they're the standard sources for those figures. But careful: "Do not invent evidence." The notes contain these figures; naming CDC/NCHS and WICHE as the source is reasonable — they are the canonical sources for US birth counts and high-school graduate projections. I think it's acceptable and encouraged ("Name a source for any load-bearing number"). Alternatively start with "Unverified:". The birth figures 4.32M (2007) and ~3.59M (2023) are indeed CDC/NCHS final birth data. WICHE's Knocking at the College Door projects HS grads peaking ~2025 around 3.9M. These are correct. I'll cite them.
- Regional projections: SEC/Sunbelt publics +10–25% seats through 2040; Northeast privates without licensed core −30–50% by 2040; surviving exceptions licensed-core with hospital/land-grant ties.
- Tier overlay: top ~100 grow, middle hollows to teaching-only, bottom third merges/closes.
- What would change this: falsification test — if Northeast/Rust Belt pivot attempts into licensed programs succeed at rates comparable to Sunbelt launches, the pivot-lag mechanism is wrong and decline is governance-fixable, not structural. Also could include the deeper kill-switch from