A group's desire for harmony overriding realistic evaluation of alternatives, leading to poor collective decisions.
Imagine you're in a car with all your friends, and the driver takes a wrong turn. You notice, but everyone else seems fine with it, so you stay quiet. Then someone else notices too, but they also stay quiet because nobody else is speaking up. Pretty soon, everyone is privately worried but publicly silent, and you all end up lost — because nobody wanted to be the one to say 'Hey, I think we're going the wrong way.'
Groupthink emerges when members of a tightly bonded group prioritize social cohesion and consensus above rigorous evaluation of evidence and alternatives. Under these conditions, individuals self-censor doubts, suppress dissent, and rationalize warnings that challenge the group's preferred course of action. Self-appointed 'mindguards' may actively shield the group from contradictory information, while an illusion of unanimity makes silence appear to be agreement. The result is a decision-making process that fails to consider risks, ignores outside expertise, neglects contingency plans, and produces outcomes far worse than any individual member would have reached alone.
The same glitch looks different depending on the terrain. Finance, medicine, a relationship, a team — same mechanism, different costume.
Investment committees in funds or banks tend to converge on a single thesis championed by the most senior or most vocal member, with analysts self-censoring contradictory findings to preserve team harmony, leading to concentrated, poorly hedged positions.
Surgical teams or diagnostic committees may defer to the attending physician's initial assessment, with nurses and junior doctors withholding observations that contradict the senior diagnosis, increasing the risk of misdiagnosis or procedural errors.
Faculty committees evaluating curricula or tenure cases may suppress individual reservations to maintain departmental unity, resulting in the approval of weak candidates or programs that no single member would have endorsed alone.
Friend groups or families may collectively avoid addressing a member's harmful behavior — such as substance abuse or a toxic relationship — because no individual wants to be the one to disrupt the group's harmony.
Engineering teams may rally around a technical architecture championed by a respected lead, with individual developers suppressing concerns about scalability or security to avoid being seen as obstructing progress, resulting in fragile systems.
Hiring panels often converge on the candidate who receives the first enthusiastic endorsement, with panel members who had concerns about the candidate declining to voice them after sensing the group's positive momentum.
Political advisory groups or editorial boards may develop shared narratives that go unchallenged internally, leading to policy decisions or coverage angles that ignore critical counterevidence because no member wants to break consensus.
The term was coined by William H. Whyte Jr. in 1952 in Fortune magazine. Irving Janis, a research psychologist at Yale University, developed the formal theory in his 1972 book 'Victims of Groupthink' and revised it in 1982.
In ancestral environments, group cohesion was literally a matter of survival. Being expelled from a tribal group meant near-certain death from predation, starvation, or rival groups. Brains that prioritized harmony over individual dissent helped maintain the cooperative bonds necessary for collective defense, hunting, and child-rearing. The cost of occasionally making a suboptimal group decision was vastly outweighed by the benefit of keeping the group intact and functioning.
AI systems can both amplify and simulate groupthink. When training data reflects consensus-driven outputs — such as articles, reports, or decisions produced under groupthink conditions — models inherit those biased conclusions as ground truth. In human-AI interaction, LLMs' tendency toward agreeableness and accommodation can create a two-person groupthink dynamic, where the AI reinforces the user's existing views rather than challenging them, mimicking the role of a conforming group member.
50 cards are free to preview. Buyers unlock the rest of the deck plus the interactive training — Spot-the-Bias Quiz unlimited, Swipe Deck with spaced repetition, My Blindspots, Decision Pre-Flight, the Printable Deck + Cheat Sheets, and the Field Guide e-book. $29.50$59.
