The most underreported AI story isn't about job destruction — it's about decision-making atrophy at the top of organizations. While executives benchmark competitors' AI adoption and announce transformation initiatives, Johnathan and Melissa Nightingale, who work directly with companies on management and leadership training, are watching something quieter and more corrosive: managers offloading judgment to AI tools and losing the capacity to lead without them.
The Gap Between Executive Hype and Office Reality
The Nightingales' vantage point is specific and credible. They aren't economists modeling labor displacement or researchers running controlled trials — they are practitioners inside organizations, watching AI adoption happen in real time. Their account, surfaced in conversation with Atlantic writer Charlie Warzel, suggests a consistent pattern: CEOs are disproportionately "AI-pilled," convinced of transformative productivity gains, while the actual experience on the ground is messier, more ambivalent, and often counterproductive.
This mirrors a broader dynamic that has played out in prior technology cycles. During the enterprise software boom of the late 1990s, executives mandated ERP implementations that consumed billions and frequently degraded operations before delivering any promised efficiency. The adoption pressure came from the top; the friction was absorbed by the middle. AI in 2024 follows a structurally similar pattern, with one important difference: the tools are cheap enough to deploy without formal implementation, which means the chaos is more diffuse and harder to audit.
The specific concern the Nightingales raise — that AI is becoming a feedback crutch — is worth dwelling on. Managers are using AI-generated performance assessments, communication drafts, and decision frameworks in place of developing their own. The skill atrophies. The AI fills the gap. The organization loses institutional judgment without realizing it, because outputs look roughly similar in the short term.
Work as Social Infrastructure, AI as Solvent
The conversation takes a less predictable turn when it connects AI adoption to the loneliness epidemic. The argument is structural rather than sentimental: work, for many people, is the primary remaining site of sustained social connection. As remote and hybrid arrangements became normalized post-2020, and as AI tools now reduce the need for human-to-human interaction even in those attenuated contexts, the social substrate of work is eroding.
This isn't a novel observation in isolation — Robert Putnam's Bowling Alone mapped the collapse of civic association over decades — but the AI vector is new. Previous automation waves displaced workers from jobs; this one risks displacing them from relationships while they remain nominally employed. A knowledge worker who routes communication through AI agents, receives AI-generated feedback, and attends fewer collaborative meetings is still employed but increasingly socially peripheral.
The productivity question the episode raises — is AI even making us more productive? — remains genuinely unresolved. The aggregate data from institutions like the National Bureau of Economic Research is mixed. Gains in narrow, well-defined tasks are real; gains at the organizational level are much harder to demonstrate. The Nightingales' ground-level observation is consistent with this: companies are adopting AI tools without clear measurement frameworks, which means the productivity narrative is largely self-reported and therefore unreliable.
What this conversation ultimately surfaces is a governance problem masquerading as a technology problem. Organizations lack the managerial infrastructure to evaluate AI adoption critically, in part because AI is simultaneously being used to substitute for that managerial capacity. The circularity is the crisis. Whether companies develop the institutional literacy to break that loop — or whether a generation of leaders simply becomes dependent on tools they cannot interrogate — is the question that matters most, and it remains wide open.
Source · The Frontier | AI
