Leadership Summit 2026: Inspiring Grit and Growth Amid Unprecedented Change
HBR.org · podcast · 2026-07-23 Reader 🖥️
Adaptive Advantage / Adaptive CapacityAI-as-Coach Effect (Learning-Environment Quality)Coaching Styles (Situational)Cognitive SurrenderGrit (Passion & Perseverance)Situational Agency (Environment Design)Systemic Apprenticeship & Skill Transmission
AI's effect on human skill and grit is not fixed by the technology itself but by whether people treat it as a coach to learn from (attending to and imitating its superior output) rather than a crutch to passively accept, and by whether they also deliberately design their surrounding situation — mentors, teams, culture — since grit (steadfast top-level goals, agile tactics) is necessary but not sufficient for excellence.
practical · mixed — A research-grounded talk (grit theory, Ericsson's learning-curve research, an original RCT) delivered as practical leadership advice: the propositions build toward 'here is what to do about AI as a leader and individual,' via lecture, live Q&A, and anecdote.
Structure
| Part | Role |
|---|---|
| Definition of grit: passion + perseverance for long-term goals, expressed as a goal hierarchy that is 'stubborn at the top, agile at the bottom' | context |
| Ericsson's learning-curve research (the '10,000-hour rule') establishing that sustained effort, not talent, keeps people improving | evidence |
| Gallup/Walton Family Foundation survey of Gen Z AI use and ambivalence | evidence establishing the stakes |
| The Lira cover-letter-rewriting RCT: AI-assisted participants showed durably improved, not diminished, unassisted writing skill | central evidence / turning point |
| The learning-environment-quality mechanism: Federer's-coach analogy and John Wooden's explanation-demonstration-imitation-repetition model, applied to AI as a demonstrator | explanatory mechanism / conclusion |
| Q&A: cognitive-offloading concern, answered via Ben Lira's personal grief-to-adaptation narrative | counterargument raised and answered |
| Q&A: maximizing vs. satisficing (Herbert Simon) as the actionable mindset distinction | refines conclusion into practical guidance |
| Q&A: AI ambivalence reframed as a rational response to real accountability gaps, not a grit deficit | qualification / limits the argument's own framing |
| Q&A: situational agency at scale, illustrated via Michael Phelps's coach/facility/team/mother | second, parallel pillar of the argument (environment design, from the forthcoming book 'Situated') |
Key terms
- Grit Grit (Passion & Perseverance)
Passion and perseverance specifically for long-term goals — not intensity, excitement, or extroversion, but consistency and loyalty to a top-level goal sustained over weeks to decades. Measured by a four-item subset of Duckworth's grit scale (two perseverance items, two reverse-scored passion items). - Goal hierarchy
The layered structure of a gritty person's motivation: a stable top-level goal ('ultimate why'), mid-level goals (3-5 year plans), and a bottom-level to-do list. Effective grit means being 'stubborn at the top' (mission/values don't change) but 'agile at the bottom' (tactics adapt to circumstances, including new tools like AI). - Learning environment (quality)
The surrounding conditions — coaching, feedback, demonstration — that determine whether a given amount of effort translates into skill gain. Duckworth's reframe of the 'does AI make you dumb' debate: effort and environment quality are separate variables, so reduced effort (from using AI) does not have to reduce learning if environment quality (AI as a superior demonstrator) rises enough to compensate. - Situational agency Situational Agency (Environment Design)
The deliberate curation of one's physical, social, and cultural surroundings (which coach, which team, which culture, where you keep your phone) to bring out one's best — treated as a lever as important as personal traits like grit. Named as the subject of Duckworth's forthcoming book, 'Situated.' - Maximizing vs. satisficing
Borrowed from Herbert Simon: satisficing means accepting 'good enough' as good enough (most of life); maximizing means treating a specific chosen domain as one where good is never good enough. Duckworth's point: whether an interaction with AI becomes a learning opportunity depends on whether the user holds a maximizing mindset toward the skill in question.
Propositions
- explicit⚡ disputed & corroboratedGrit is passion and perseverance for long-term goals — consistency and loyalty to a top-level goal sustained over months, years, or decades — not intensity, excitement, or emotional energy.conf: highsupports ←Building an Autonomous Enterprise for Real-World Services with Netic Founder Melisa Tokmak · p10 — Her critique of short-termist '18-month' thinking, and her own multi-year commitment narrative, match the grit definition's emphasis on sustained consistency over intensity or urgency.contradicts ←Tiny Experiments · p1 — Le Cunff's thesis (fixed linear goals constrain growth; favor small time-bound experiments over a fixed 'tyranny of purpose') is in tension with Duckworth's grit construct, which defines high achievement as sustained passion and loyalty to one top-level goal over months to decades — the opposite prescription for how to relate to long-term commitment.
- explicitGrit is necessary but not sufficient for excellence; high achievers are also embedded in superior coaching, teams, and environments.conf: high
- explicitA gritty person's goal structure should be stable at the top (mission/values don't change) but agile at the bottom (day-to-day tactics adapt, including adopting new tools like AI), and this agility is what lets grit and AI adoption coexist.conf: high
- explicitdevelopedStaying on a learning curve through sustained effort matters more than raw talent for reaching world-class performance (per Ericsson's research on chess masters, ballerinas, and CEOs).conf: highsupportsHow to Build a Superteam That Keeps Getting Better · p1 — Both converge on learning speed/capacity, not existing talent or plans, as the decisive performance variable — one at the individual level (grit + learning curve) and one at the team level (superteams).extends ←The Laws of this World (Game Theory) · p5 — Same Ericsson deliberate-practice research; this essay adds the condition that effort must be aimed at a well-selected problem, refining the leadership-summit doc's more general 'sustained effort beats raw talent' claim with the feedback/problem-selection mechanism Ericsson's own work specifies.
- explicitThree-quarters of surveyed Gen Z adults (18-28) report using AI in the past month, primarily for productivity, but feel deep ambivalence — fearing AI use will make them lazier and less smart over the long term.conf: highsupportsThe Psychological Costs of Adopting AI · p10 — Duckworth's Gen Z survey finding (deep ambivalence, fear of becoming lazy/dumb) empirically converges with the psychological-debt research finding that early-career workers report the highest cognitive and identity debt from AI use.
- explicitContrary to both experts' and participants' own predictions, people who used an AI tool to instantly rewrite a bad cover letter showed durably improved (not diminished) unassisted writing skill a day and a week later — better than people who practiced alone or did no practice at all.conf: high
- explicitUsing AI measurably reduces effort during a task (fewer keystrokes, less time spent), and reduced effort by itself should be expected to reduce learning.conf: highrefinesOur Favorite Management Tips on Giving Feedback · p6 — Duckworth's finding that reduced effort doesn't necessarily impair learning — provided the learner watches and imitates a superior demonstration — complicates the simpler 'growth requires struggle, let them struggle before stepping in' claim by adding observational learning as a substitute channel.
- explicitSkill improved despite reduced effort because AI raised the quality of the learning environment: when AI produces better output than the user could alone, it functions like a great coach demonstrating a skill, and the user learns by watching and imitating the example — echoing John Wooden's 'explanation, demonstration, imitation, repetition.'conf: highextendsCognitive Surrender · p7 — Duckworth's learning-environment-quality mechanism gives a concrete causal account of exactly what 'posture' (thinking with vs. instead of thinking) cashes out to: attentive, imitative watching of a superior AI demonstration, rather than passive acceptance.
- explicitIt is rational to delegate to AI only the tasks at the bottom of one's goal hierarchy — tedious, low-learning-curve chores — while continuing to invest personal effort and attention where learning and expertise-building actually matter.conf: high
- explicitWhether AI use makes someone lazier and less capable, or sharper, is substantially a matter of individual choice/posture — specifically, whether the person holds a 'maximizing' mindset (a domain where good is never good enough) toward the skill in question, rather than passively accepting AI output everywhere.conf: high
- explicitThe human brain and body are adaptive/plastic in both directions — capable of atrophy as well as growth — so cognitive offloading to AI carries a genuine, not merely hypothetical, risk of skill and grit loss.conf: high
- explicitYoung workers' ambivalence and hesitation about adopting AI is often a rational response to genuine risks — occasional AI errors/hallucinations and real accountability gaps — not simply a personal deficit of grit or resistance to change.conf: high
- implicitDespite that hesitation being legitimate, leaders and individuals have no real option to disengage from AI: declining to ask how it could improve their work is itself a costly decision.conf: medium
- explicitSustained individual excellence depends on 'situational agency' — deliberately curating one's coach, teammates, physical environment, and culture — not on willpower or personal grit alone.conf: highsupportsHow to Turn Individual Talent into Organizational Excellence · p9 — Both cite Olympic swimmers to argue sustained excellence is produced by systemic/environmental design (coach, team, routine, culture) rather than individual brilliance alone.
- explicitGrit is not fixed; it changes across the lifespan, and on average people become grittier with age and experience, meaning grit can be deliberately built once its underlying goal-hierarchy structure is understood.conf: high
Arguments
AI use can durably increase a person's own skill rather than erode it, when reduced effort is offset by a higher-quality learning environment. strength: medium
Explicit premises:
- In the cover-letter RCT, AI-assisted participants' unassisted writing skill improved (not declined) a day and a week later, beating both solo-practice and no-practice conditions.
- AI use measurably reduces effort (fewer keystrokes, less time).
- When AI's output is better than what the user could produce alone, the user can learn from it by watching and imitating, the way a student learns from a coach's demonstration.
Implicit premises:
- Skill transfer from watching AI's output requires the learner to actively attend to and imitate the demonstrated output, not simply accept it and move on.
- The effect found in a lab cover-letter-rewriting task generalizes to other skill domains and to sustained, real-world AI use over months or years, not just a single short exercise.
Inferences:
- If learning-by-example from AI output generalizes, organizations could deliberately design AI interactions to surface the AI's reasoning (not just its final answer) as a systematic upskilling mechanism.
Grit is necessary but not sufficient for excellence; situational factors (coaching, team, physical environment, culture) are jointly required. strength: medium
Explicit premises:
- Michael Phelps had elite coach Bob Bowman for 20 years.
- Phelps trained 10 minutes from the premier aquatic youth facility in the world.
- Phelps was on a team that broke records and went to the Olympics routinely.
- Phelps's mother, Debbie Phelps, built a 'culture of excellence' at home.
Implicit premises:
- These situational advantages in Phelps's biography are causally responsible for his excellence, not merely correlated with grit that would have produced similar results regardless of environment.
- Grit and situational advantage are separable, independently necessary factors, rather than situational advantage being itself largely a product of the individual's own gritty choices.
Inferences:
- Leaders cannot rely on selecting or cultivating 'gritty' individuals alone; they must also deliberately engineer team and environmental conditions.
Narratives
- A hypothetical Roger Federer with a world-class tennis coach versus the same Federer with a terrible coach who barely plays tennis and knows nothing about motivation.
→ Illustrates that AI's effect on learning is analogous to a coach's effect — the quality of the 'coach' (human or AI) determines whether reduced independent effort helps or degrades development. - Ben Lira (the graduate student who led the cover-letter RCT) went through despair upon realizing GPT could out-code him after he'd spent his 20s learning to program, then adapted by delegating only tedious formatting tasks to AI while learning from its code by example and freeing time for higher-value work.
→ Concretizes the abstract argument (delegate the bottom of the goal hierarchy, learn from the top) with a first-person account of converting cognitive-offloading anxiety into productive practice. - Michael Phelps's biography: coach Bob Bowman from age 11 to 31, a childhood 10 minutes from an elite aquatic facility, a record-breaking team, and mother Debbie Phelps's 'culture of excellence.'
→ Supports the situational-agency thesis that sustained excellence is a deliberately designed situation, not inner grit alone.
Questions addressed
- ✅ Does grit still matter, or even help, in a world where AI can do much of our thinking for us?
Yes — grit still matters because it keeps a person on a learning curve, but its expression must include agility about incorporating tools like AI at the tactical, bottom-of-hierarchy level. - ✅ Does using AI to complete a task make a person's own underlying skill worse?
Contingent: reduced effort alone predicts worse skill, but if AI models excellence and the user treats it as a demonstration to learn from, the improved learning-environment quality can outweigh the effort reduction and produce durable skill gains. - ✅ Is young workers' ambivalence about AI a symptom of insufficient grit or resistance to change?
No — Duckworth explicitly disputes this framing, arguing the ambivalence is a rational response to AI's genuine error rate and real accountability gaps, not a personal-trait deficiency. - ⭕ How does 'situational agency' (designing environments rather than relying on willpower) apply when leaders try to scale it across a large organization, and where does the framework break down?
Not actually answered — Duckworth responds with more individual-level biographical detail about Phelps's coach, facility, team, and mother rather than addressing the organizational-scale mechanism or failure modes the question specifically asked about.
Gaps
- incompleteThe specific question asked — how situational agency applies 'at scale across a big organization' and where the framework breaks down — is never actually answered; Duckworth reverts to the individual-level Phelps story rather than addressing organizational design, roles, or failure modes.
Read next: Read the corpus's 'Individual Talent, Organizational Excellence' HBR piece for the systems-level, cross-role answer to this exact scaling question (talent + team + routine, with responsibility distributed by role). - incompleteThe cover-letter RCT is a single, narrow task (short-form business writing) under lab conditions, with unspecified sample size and no report of how the 'watch and learn' mechanism was verified — there's no manipulation check on how much participants actually attended to the AI's output versus simply accepted it. It's unclear whether the effect generalizes to more complex, longer-horizon skills (e.g., coding, strategic judgment) or requires a specific attentional posture that most real-world AI users won't adopt without explicit prompting.
Read next: Read the underlying Lira/Duckworth study directly; cross-reference with the corpus's claim that attention (not prompting skill) is the real bottleneck on AI competence, and with the cognitive-surrender literature's opposite finding (surface-correctness signals causing surrender) in a professional coding context rather than a lab writing task. - illogicalThe Phelps anecdote is offered as evidence that environment (not just personal trait) shapes outcomes, but Phelps's own choices — following coach Bob Bowman when he relocated, 'navigating to be around the best teammates' — are themselves described as gritty acts of agency. This blurs the very trait/situation distinction the argument is meant to establish: any environmental advantage can be re-described as a product of the individual's own grit in seeking it out, risking an unfalsifiable model.
Read next: Look for research that operationalizes situational agency independently of grit — e.g., through random or exogenous assignment to environments (coach changes, team reassignments) rather than a single retrospective case study, which cannot disentangle the two. - incompleteNo discussion of who bears responsibility, or what safeguards apply, when the 'learn by watching AI' mechanism fails — e.g., in domains where surface-correctness signals or an agreeable AI tone cause people to accept wrong output without noticing, which is precisely the failure mode the optimistic learning-environment mechanism assumes away.
Read next: Cross-reference with the corpus's cognitive-surrender extraction, which documents that 73% of participants accepted a wrong AI answer with rising confidence regardless of correctness — a direct counter-case to the 'AI as reliable demonstrator/coach' assumption.
Closes gaps elsewhere
- Brain Food: Answers Without Effort · g1 — Duckworth's cover-letter RCT distinguishes effort from learning-environment quality as separable causal levers, offering exactly the requested mechanism: reduced effort does not automatically reduce learning if environment quality (via a superior demonstration to watch and imitate) rises enough to compensate — distinguishing productive observational learning from mere toil-reduction.
- Brain Food: Two Types of Talent · g1 — Duckworth names the missing mechanism directly: 'chosen talent' operationalizes as grit's goal-hierarchy structure (steadfast top-level goal, agile tactics) combined with deliberately curated situational agency (coach, team, environment) — not effort alone, and not innate ceilings vs. effort as a binary.
- 11) AI-spesial med Vilde Opsal og Anders Sveen · g2 — Duckworth offers a partial mechanism: delegate only the genuinely low-learning-curve, tedious tasks at the bottom of the goal hierarchy to AI, while treating AI's output on higher-value tasks as a demonstration to actively watch and imitate (explanation-demonstration-imitation-repetition), rather than assuming all automated drill-work is equally lost as a training opportunity.
- SMOC.AI's Post — 5 Minutter: AI og utviklere · g4 — The Lira RCT is direct empirical evidence for this gap's hypothesis: participants who watched AI produce superior output (rather than doing the task fully themselves) showed durably improved unassisted skill, supporting the claim that watching/reviewing can be as powerful a learning mechanism as first-hand doing.
Related documents
- How to Build a Superteam That Keeps Getting Better
- Our Favorite Management Tips on Giving Feedback
- The Psychological Costs of Adopting AI
- How to Turn Individual Talent into Organizational Excellence
- Cognitive Surrender
- Teaching Coding When AI Can Write the Code
- Design AI Systems That Actually Strengthen Human Reasoning
- Why Mentoring Matters More in the AI Era
- Research: Why Some Junior Employees Work Well with AI—and Others Don’t