The Gap Between Knowing and Doing
The gap between knowing and doing names the structural feature of human moral psychology that traditional ethics education has systematically failed to address. Gentile's research across industries and decades produced a consistent finding: the professionals who fail ethically are not people who lacked principles. They are people who could often have written the ethics textbook they violated. What they lacked was the practical skill to translate principles into action under real-world pressure — the rehearsed scripts, the anticipated objections, the coordinated peer support, and the institutional conditions that convert private conviction into public voice. The gap is Aristotle's akrasia — knowing the good and doing otherwise — given operational precision. It is not a failure of character but a failure of preparation, and the distinction changes what counts as an adequate intervention.
In the AI Story
The gap's diagnostic power lies in what it rules out as explanation. When ethical failures occur, the conventional response is to demand more ethics training — more frameworks, more principles, more awareness. The response assumes that failure follows ignorance. Gentile's evidence shows that this assumption is almost always wrong. The training produces articulate analysts who can identify dilemmas with precision and navigate them with no improvement in actual behavior. Thilo Hagendorff's 2020 evaluation of AI ethics initiatives replicated the pattern: developers who had read comprehensive guidelines made the same decisions as developers who had not.
The gap is particularly visible in the COMPAS case, where engineers and data scientists inside Northpointe had access to the same data that ProPublica would eventually analyze. The differential error rates across racial groups were visible in the outputs to anyone who examined them. The case study Gentile built around this situation does not ask whether anyone saw the pattern. It asks what the person who saw it said, to whom, and with what preparation — forcing students to confront the performance question the awareness framework never posed.
The Orange Pill's account of AI-era moral failure — the builder who cannot stop at three in the morning, the senior architect who confesses grief only in a hallway, the engineer who oscillates between excitement and terror without speaking — describes the gap with unusual honesty. Each figure knew. Each did not act on the knowing. Gentile's framework would treat none of them as deficient in character; she would treat all of them as professionals who had never been taught how to perform what they were already aware of.
The AI transition's temporal compression makes the gap's consequences more severe than they were in slower-moving industries. The decision window opens and closes in weeks. The unprepared professional who waits until the moment of decision to formulate her ethical argument will find the moment has passed. Temporal compression transforms preparation from a professional luxury into a structural necessity.
Origin
The gap emerged as an explicit concept through Gentile's post-Harvard work, when she began interviewing executives and mid-career professionals about ethical decisions in their actual organizational lives. The pattern was immediate and consistent. When asked to describe situations in which they had acted against their values, respondents almost never reported confusion about what was right. They reported knowing and being unable to act — a phenomenon so common that Gentile began treating it as the primary phenomenon ethics education needed to address, rather than the outlier it had been assumed to be.
Key Ideas
Knowledge is widely distributed; action is not. Ethical failures cluster not in populations that lack principles but in populations that lack preparation.
Aristotelian akrasia is operational, not metaphysical. The phenomenon of knowing the good and doing otherwise has specific, addressable causes — rehearsal deficits, peer isolation, institutional discouragement — rather than an inherent weakness of will.
The gap is systematic, not individual. It appears reliably across industries, cultures, and organizational contexts, which means it is produced by structural conditions rather than personal failings.
The gap's remedy is specific practice. Generic exhortation to act ethically is exactly as effective as generic exhortation to be brave or competent — that is, not effective at all.
Appears in the Orange Pill Cycle
Further reading
- Aristotle, Nicomachean Ethics, Book VII (on akrasia)
- Mary Gentile, 'Ways of Thinking About Our Values in the Workplace' (GVV baseline thought experiment)
- Thilo Hagendorff, 'The Ethics of AI Ethics: An Evaluation of Guidelines' (2020)
- Lynne Paine, 'Managing for Organizational Integrity' (Harvard Business Review, 1994)