Does Education Matter in the Age of AI? MIT President Sally Kornbluth
In a Nutshell
MIT President Sally Kornbluth emphasizes sustaining meritocracy and excellence at scale by prioritizing merit-based hiring, rejecting mediocrity ("If you take a lick of the lollipop of mediocrity, you will suck forever"), and maintaining strict standards like MIT's needs-blind admissions and reinstated SAT/ACT requirements, even against favoritism or external pressures. She shares crisis leadership lessons from her congressional testimony and MIT's rejection of the federal Compact for Excellence, stressing calm external demeanor, strong board support, and clear communication. In the AI era, education evolves to teach critical thinking, AI evaluation, and human-AI collaboration beyond rote skills, with MIT fostering interdisciplinary innovation and local ecosystems for real-world impact.
These notes were generated by AI and may contain inaccuracies.
The way to maintain meritocracy and excellence is to ensure every faculty, staff, and student hire focuses on excellence. A colleague at Duke had a sign: > "If you take a lick of the lollipop of mediocrity, you will suck forever."
Sally Kornbluth is president of MIT. Within a year of being hired, she testified in Congress alongside the presidents of Harvard and UPenn. Elise Stefanik asked tough questions; all three presidents struggled. Kornbluth's response was excellent. The Harvard and UPenn presidents did not survive the crisis; Kornbluth thrived. She recently responded to the federal government's Compact for Excellence in Higher Education, one of 11 institutions selected. MIT's response showed terrific leadership.
MIT, over 150 years old, sustains excellence and meritocracy. Natural tendency is regression to the mean. Discussion on scaling meritocracy for CEOs.
"If you take a lick of the lollipop of mediocrity, you suck forever."
Topics include CEO blocking and tackling, hiring and managing a board.
Nephew Jack applying to Sloan; host tried influencing via every door, but MIT admissions indifferent: > "We don't care who you are. You're not going to influence the process."
MIT first to reinstate SAT and ACT post-COVID, needs-blind admissions.
MIT admissions pure merit; no special favors, even from president. To scale meritocracy: consistently focus on excellence for every hire. Better to have nobody than mediocre hires.
When starting a company or team, founder is best; temptation to hire just anyone. Maintain excellence so collaborators know everyone does good work.
Consistency across institution critical; one open door would flood with favoritism. Nephew got in on merit; everyone assumes influence, but none used.
Merit multidimensional: high grades and test scores needed, but also ability to benefit from and contribute to MIT environment. Not all quantifiable; flexibility for various excellent people. Criteria in mind but not rigidly written; "if you know it, you see it."
Around 150 employees, founder stops interviewing everyone; bar drops, noticeable at 500. Advice: Easier to prevent slide than recover. Build strong team of trusted lieutenants early; delegate without micromanaging after verifying. Convey standards down hierarchy.
Even at MIT, communicating messages challenging. Example from prior institution post-COVID: reopening email had 30% open rate. Need high-touch messengers to fan out personally.
Most CEOs have trust issues, trust few for cross-functional hires. Kornbluth: No trust issues; assesses capabilities before decisions. Be frank immediately if underperforming; don't wait as personal connections grow. Tell people clearly when doing poorly or well. People crave hearing praise.
CEO trope: Praise 5 times per correction. Kornbluth agrees if genuine; costs nothing, shows gratitude. Don't fake praise to temporize.
Staff feedback: Kornbluth "bold but not overbearing."
Scientist training grad students: Labs have highs (exciting) and lows ("dark tea time of the soul" when experiments fail). Motivate by understanding individual needs and motivations; cater to what makes them productive and happy. Doesn't scale infinitely, but key for immediate team.
Columnist George Will called Kornbluth "inexplicably ebullient." CEO days: meetings, bad news inbox, annoyance. Kornbluth: Jaw-dropping discoveries daily; enabling science advancement better than own work. Find humor; can't make up some events.
Biology: Receptor downregulation – constant stimulation desensitizes. After years of incoming, choose pleasure in cool work over misery. Gets pissed daily but rises above.
Gets down when personal, e.g., post-Congress testimony on antisemitism: Called antisemite despite being Jewish; personal emails offensive. Transiently down; deals by talking incessantly, wearing on sleeve.
Summoned to Congress shortly after starting; not optional (subpoena threat). Consulted team including VP Communications Alfred Ironside, university friends. Naive expectation of real dialogue; was political theater. Political science major disappointed in quality.
Prepared well; full 4-5 hour testimony fine until gotcha moment. First on key question (Harvard/Penn first on others). Should have said: > "Oh my god, nobody would say that on our campus." Hindsight 2020; factual like Dukakis moment.
Realized issue post-hearing via newspapers, texts from Duke friends, Saturday Night Live.
Leaders must appear calm/measured externally despite internal reactions. Felt unfair accusation of views not held.
Aftermath: Clarified remarks quickly to community without creating news cycles or TV appearances. Pivotal: Board (Corporation) supported publicly early, enabling focus on MIT work. Many board members listened to full testimony/transcript, defended to critics: > "Did you listen to the whole thing?"
Key takeaways: Stay calm, strong board backing invaluable, clarify thoughtfully.
Compact to 11 institutions; MIT first to respond, politely declining pre-deadline.
Immediate reaction: No way to sign; violates scientific merit for funding (prefers behaviors/commitments). Called chair Mark Orberg: > "I'm not signing this." Full board supported.
MIT's merit story: No legacy/development admissions, free tuition under $200k family income, reinstated SATs first. Story told itself; tone "thanks but no thanks" – compete on merit.
Letter by writer Martha Edison; long draft iterated by four or five over days.
Collaborative process for writing MIT's compact letter involved pieces written by Sally Kornbluth herself, various team members, and final words refined over extra days with the executive committee of the board. They argued over phrasing until EVP Glenn Shore suggested a key phrase that finalized it. Emphasized group thinking, team cohesion allowing bluntness. Example: Martha drafts something, Kornbluth says it's not right, Martha adjusts. Brought up board more than expected.
Advice on Building and Managing Boards
All CEOs must build boards before going public. MIT has a very large board of 75 people, unusual for most companies, with a core executive committee. Seek range of expertise, but fundamentally smart people with integrity open to arguments. Avoid those with strong views unwilling to consider others—frustrating. Unusual to sync easily with board; friends have varying board experiences, some requiring constant maneuvering. Ideal: governance but not management, allowing Kornbluth to do her job. Don't agree on everything—e.g., many phone calls on compact letter saying no to certain ideas. Critical: open exchange, honesty without parsing words. Board calm in crises; many are or were CEOs, navigated high-stakes situations, understand staying calm.
MIT extremely decentralized. Deciding to centralize edge activities is thornier than choosing initiatives. In budget crises, eliminating duplication hard—everyone sees their research program or team as personal babies. Centralize only when necessary for efficiency, financial reasons, or better function, then navigate resistance until benefits clear. Initiatives often bottom-up: spot great work and capture it. Presidential initiatives encourage interdisciplinarity—high-level collaborators yield good partners.
Handling Prima Donnas
Every CEO asks about prima donnas (brilliant assholes). MIT has many, but most have the goods—not like those who think they're great when not. If Nobel winner defends program, no argument. Makes taking things away harder. To take away: engage, provide good rationale showing thought process—not personal dislike. Most accept; some hate her for it, but generally understood as necessary. Workforce largely tenured, different from corporate. Say yes sometimes, but no huge amount of time. On founder mode (e.g., Jack Dorsey laying off 40%): centralized micromanagement in style, but MIT context unique.
Criticism of higher ed: Thiel Fellowships, Marc Andreessen advising against college, Alex Karp's Palantir Fellowship. With AI, what does education look like in five years? College experience beyond skills—alums say MIT changed lives via environment, not specific class. Pedigree, learning, network valuable; skills learnable anywhere. Train humans to use AI as tool augmenting performance. Uncertainty: how much to memorize for creativity vs. offload to AI. AI gives different answers to same question phrased three ways—teach evaluating that, spotting hallucinations. Students need basic coding to verify AI output, conceptualize questions. Writing is thinking—not same as AI drafting; use AI to iterate.
Evolving Pedagogy and Physical AI Limits
MIT focuses on effective AI tool use. Building things core; physical AI lags (e.g., robot spilling Coke can). Classrooms may evolve to Oxbridge-style tutorials: students work with AI tutor in groups, then discuss with faculty. Traditional didactic lectures may not survive with factual alternatives, but human interaction, discussions, critiques essential.
For hypothetical 17-year-old daughter: has 30-year-old surgeon daughter—good field, though parts automated; she's into robotics. If 17 herself (biologist), pick neuroscience or immunology—mysteries of brain, health impacts, years of discovery. Only pick passion areas as thinker. Grad school past: PhD sequencing single gene was technical, not engaging. Now, science discovery companies give great students AI 'hands,' but they must be creative thinkers. Don't do PhD as 'hands'—could change fast, like Google outsourcing info, AI outsourcing cognition. Even physician fields: AI changes image analysis—radiologist or pathologist? Human judgment still needed (e.g., cancer detection). Want curious students who think critically; 17-year-olds don't pick lifelong careers—people pivot.
Graduate Outcomes and Excellence
Doesn't care vocational paths (CS grads to quant shops, PhDs, startups—wishes more startups). In her biology lab, half to industry, half academia—controversial then among faculty pushing academia. Doesn't care destination if prepared to excellence. No tailoring to specific goals; provide knowledge base to do anything. No cutting corners.
KATE initiative to rapidly bring student/faculty discoveries to world, populate local ecosystem. Biotech impacted area; AI going West Coast. Economic reasons for region/country. MIT part of ecosystem—students/faculty back-and-forth. Continuing dialogue between MIT and real world needs proximity. More co-ops in local companies require proximity. Massachusetts vitality from education + cutting-edge tech economy.
Host's first exposure: congressional hearing—initially low NPS, but admired crisis handling (hecklers at graduation). Proud MIT alum. Takeaways: Handled whopper crises well (like Vlad from Robinhood). In tornado crisis, show calm exterior—unnatural but essential. Mistake (e.g., Congress): Admit quickly, clarify; when explaining, you lose. Have excellent writer on team—well-written posts gold internally/externally (HubSpot example; host took writing classes, leaned on co-founder).
Board, Meritocracy, Hiring Advice
Good board behind you underrated. Sustaining excellence/meritocracy: HubSpot imported MIT culture. Referrals (e.g., host's mom) didn't help—merit-based, irritated family but good. Interviewing: direct reports always; up to 100 employees, do final 10-minute veto interviews. HubSpot values: post-interview form ranked 1-4 on all five, tracked. Hire spiky people (weaknesses but great per some) over safe liked-by-all—paid off (e.g., Yamini Ramineni, current CEO).
Keep Sequoia Capital in your library
Save the videos and channels worth coming back to, and find them again in one place.
