“Do the right thing even when it’s hard.”
Mary Barra used this principle while rebuilding trust at General Motors after the ignition-switch crisis. The words are deliberately plain because ethical failure often survives inside complicated procedures, divided responsibility, and pressure to protect short-term results. “The right thing” begins with safety, truth, and accountability; “even when it’s hard” acknowledges that these duties can conflict with schedules, budgets, hierarchy, and reputation. Barra’s formulation makes integrity a decision rule for moments when the cost of acting correctly is visible and the cost of delay is easier to hide.
A company can operationalize the sentence before a crisis. Employees need clear stop-work authority, protected reporting channels, rapid escalation, and leaders who reward the early disclosure of bad news. Decision records should identify who owns the risk and what evidence would trigger a pause or recall. Compensation and deadlines must not punish people for raising material concerns. When tradeoffs involve safety or rights, an independent review can prevent the team most invested in launch from judging its own readiness. Values gain force through these mechanisms, not through repetition alone.
Moral confidence can become dangerous if one leader assumes personal certainty is enough. Hard decisions contain incomplete facts and competing legitimate interests. Doing right therefore requires expertise, listening, transparent reasoning, and a willingness to revise. It also requires distinguishing an ethical boundary from a preference dressed in moral language. Organizations should name nonnegotiable duties, consult people exposed to the consequences, and preserve an appeal path. Integrity is stronger when it can withstand scrutiny than when it depends on private conviction.
The principle applies directly to AI and other systems whose harms may emerge after release. A model can meet a performance target while creating unsafe edge cases, unequal outcomes, or misleading confidence. Teams can run red-team tests, document limitations, monitor real-world effects, and define when a product will be restricted or withdrawn. Leaders should communicate what failed and how affected people will be made whole. Barra’s sentence provides a cultural threshold: when evidence reveals a serious risk, will the organization absorb inconvenience now rather than transfer a larger cost to customers later? Incident reviews should separate learning from discipline while refusing to blur accountability. The team reconstructs what information was available, which incentives shaped attention, where escalation failed, and what safeguard will change. Individual conduct is addressed through a fair process outside the learning conversation when necessary. This separation encourages complete evidence without creating impunity. It also prevents a public promise of integrity from ending with one person blamed while the conditions that made the failure predictable remain intact.
Barra used the principle while leading General Motors through a safety crisis and a broader integrity reset.
Its simplicity directs attention to truth, safety, and accountability when organizational incentives encourage delay.
Stop-work authority, protected reporting, independent review, and clear risk ownership make integrity actionable.
Ethical confidence still needs evidence, consultation, transparent reasoning, and a way to challenge mistaken judgment.
AI systems may meet aggregate targets while hiding harmful edge cases or unequal outcomes.
Testing, monitoring, disclosure, and predefined withdrawal thresholds ensure customers do not carry avoidable corporate risk.
Explore more "Quotes of The Day"
Discover more notable quotes from influential voices across politics, science, business, technology, sports, and culture. Each quote offers insight into how ideas, beliefs, and decisions shape the world around us.
