AI 时代美国名校正在发生巨变

The familiar playbook for getting into a top US university — chase rankings, pick a major sophomore year, look for internships junior year — is breaking down. In the AI era, three forces are reshaping the entire system at once: universities are shrinking, academic programs are being rebuilt, and admissions itself is more competitive and less predictable than ever. Here's the full recap of today's session.

1 · The big picture: universities are under pressure

This is structural, not cyclical: federal research funding is being cut, state appropriations are flat or shrinking, and the demographic cliff is hitting the traditional applicant pool. Even Harvard, Stanford, Johns Hopkins, and Penn are freezing hiring, pausing PhD admissions in some humanities programs, and reallocating budgets toward AI, biomedicine, and engineering.

Picking a college is no longer just about rankings. It's about which institutions can credibly fund the next decade.

2 · Program reshuffle: what's getting cut, what's getting funded

Over the past three years, more than 1,500 academic programs have been eliminated across US universities. Syracuse restructured 84 programs. The Oklahoma State system consolidated 70. UChicago paused humanities PhD admissions in several departments. Meanwhile, AI-related programs are exploding — 193 new undergraduate AI majors launched in 2025 alone (+114% year over year), and 304 universities now offer an AI/ML degree of some kind.

The signal for families: look at where a school is investing for the next five years, not where it ranked last year. Students in programs that get cut mid-degree lose their lab access, their advisor pipeline, and the alumni network that would have helped them after graduation.

3 · The boutique advantage: small schools and consortia are winning

Small, focused colleges — Olin, Harvey Mudd, Minerva, Pomona — are outperforming on outcomes because they iterate faster, run smaller classes, and partner with bigger neighbors. The Claremont Consortium (7Cs), Five College Consortium, and the Quaker / Tri-College group let students enroll across multiple campuses, sharing 2,000–7,000 cross-registration courses without paying extra.

Concrete proof: Olin engineering grads land at Google, Microsoft, and SpaceX at a 40%+ rate. Harvey Mudd has the highest average starting salary in the US for a bachelor's degree (over $90K). And 3+2 dual-degree programs (3 years at a liberal arts college, 2 years at Columbia / Dartmouth Thayer / WashU / Caltech / Georgia Tech) let a student walk out with both a BA and a BS in 5 years.

Three questions to ask on every campus visit

1. Which schools do you have cross-registration agreements with?
2. Has this major added a dual-degree option in the last 3 years?
3. What percent of your graduates enter AI, biotech, or finance?

4 · Class of 2030: the admissions reset

Early Decision is the single biggest lever

ED/EA acceptance rates are 2–3× the regular round. Vanderbilt admits 24% ED, Duke 21%, Dartmouth 20%. At Brown and Columbia, more than half of every incoming class is locked in by November. If a student has a clear first-choice school, applying ED by November 1 is the highest-leverage move available.

Test-optional is over

Harvard, Yale, Dartmouth, Brown reinstated mandatory SAT/ACT for Fall 2025. Cornell and Stanford join them in Fall 2026. The key data point: 43% of applicants submit no scores, but only 33% of admitted students submit no scores. A 1450+ SAT or 32+ ACT is still a measurable advantage even where it's "optional."

Legacy is being legislated away

Colorado banned legacy preference in 2021. Virginia, Maryland, and California followed. CT, MA, and NY have bills in progress. The share of schools still considering legacy fell from 49% in 2015 to 24% in 2025. First-generation, low-income, Black, and Southwest-region applicant pools all grew 6–8% year over year.

AI in your application — what's allowed, what's not

Half of admissions offices now use AI to assist with file review. UNC has done so since 2019; Virginia Tech confirmed it for the 2025–26 cycle. Two-thirds of schools use AI detection tools without publicly disclosing it.

• OK with AI: brainstorming essay topics, grammar/style checks, researching schools and programs, practicing interview questions.
• Not OK: generating your Personal Statement or "Why" essay end-to-end, ghost-writing recommendation letters. Stanford, MIT, and Caltech have publicly warned about detection. If the application asks you to attest, answer honestly.

5 · Action plan: what your student should be doing now

9th grade · Class of 2029–2030

• Fall: try 4–6 activities to find direction. Winter: lock in GPA 3.7+, pick 1–2 activities to commit to.
• Spring: start one long-term project. Summer: stable college pre-college program or real summer job.

10th grade · Class of 2028

• Fall: first PSAT diagnostic. Winter: start 2–3 AP courses. Spring: confirm your "Spike" direction.
• Summer: first real research project, or a structured mentorship (Polygence / Pioneer / Lumiere).

11th grade · Class of 2027 ★ most critical year

• Fall: SAT/ACT first sitting, target 1450+. Winter: identify recommenders.
• Spring: formal rec letter requests, college visits, build a 20-school list.
• Summer: Personal Statement first draft; finish capstone research, summer school, or industry program.

12th grade · Class of 2026

• August: Common App opens — lock in your ED choice.
• November 1: ED / EA deadlines + financial aid forms (CSS / FAFSA).
• December: ED decisions — prepare RD list. Jan–Apr: RD decisions. April 1: final commitment.

11th grade is the watershed. Roughly 80% of a top-school admit comes down to what was already done before senior year started.

The 2026 admissions officer reads for six things

1. Coherent spike — one core direction + 2–4 complementary activities, not a 10-item generalist list.
2. Authentic voice — specific cadence, real detail, even contradiction. AI-flat writing is now easy to spot.
3. Demonstrated impact — what did you actually produce? Papers, products, users served, dollars raised.
4. Intellectual curiosity — self-directed research, long projects, cross-disciplinary exploration (not just competition trophies).
5. Resilience — how you respond to failure, pivots, and an unstable environment.
6. Fit & "Why-school" — specific professors, specific courses, specific clubs, specific industry corridor.

Five red flags before you apply

1. Total enrollment under 2,000 and trending down — highest closure risk.
2. Major layoffs announced in the last 2 years — clear sign of financial pressure.
3. Three consecutive years of operating losses (check Forbes College Financial Grades).
4. Your intended major is flagged "under review" — could be cut before you graduate.
5. No tenure-track hiring in 3+ years — teaching quality erodes quietly.

Quick check: Google "[school name] program review 2025" or "academic prioritization".

The three things, in one breath

1. Pick the right school. Avoid programs that are shrinking. Pick schools with an industry corridor, stable finances, and active AI-era investment.
2. Go deep on one thing. Core spike + real research output + measurable impact. It's not the count, it's the recognizability.
3. Tell a real story. In an AI-saturated application pile, your specific detail, your real voice, and your contradictions are the moat.

What's scarce now isn't "standardized excellence." It's a student with direction, drive, and long-term growth potential. Build that, and the rest of the admissions game gets a lot more navigable.