AI angst mutates into ‘FOBO’ as Fear of Becoming Obsolete takes over American workforces.

Source: Fortune

Translation: The fear is rational, the timeline is longer than the headlines suggest, and the workers most at risk are not the ones who never learned AI. They’re the ones who were excellent before it arrived and are refusing to update.

Four in ten workers now name AI-driven job loss as a primary fear; nearly double the proportion from a year ago. MIT FutureTech’s most comprehensive empirical study of AI task performance to date found that frontier AI models already complete 50-75% of text-based labor market tasks at a quality level a manager would accept without edits. That is not a projection. That is the current baseline. Failure rates are halving every two to three years, translating to 15-16 percentage point annual gains in success rates. By 2029, on the upper-bound trajectory the researchers model, AI could complete most text-based tasks at 80-95% success rates.

The MIT study’s core contribution is the shape of the disruption, not just its scale. The researchers find evidence of a rising tide (broad, incremental improvement across nearly all task types simultaneously) rather than crashing waves that suddenly master specific domains and leave everything else untouched. That distinction matters because it determines how much warning workers have. The rising tide gives you time to move. The question the data raises is whether the institutions and individuals that need to move are moving.

Most are not. Fewer than 19% of US establishments have adopted AI. Only one-third of workers say their employer is providing adequate AI training. Meanwhile, workers who do use AI are recapturing 40-60 minutes per day, and 75% report completing tasks they previously could not do at all. The companies using AI are pulling ahead. The gap between them and those that are not is compounding daily.

The primary distortion in the FOBO conversation is the frame of suddenness. Dario Amodei’s prediction that AI could eliminate 50% of entry-level white-collar positions within five years, Senator Warner’s projection of 35% new college graduate unemployment within two years, and the general tenor of executive anxiety all imply a disruption that arrives before adaptation is possible. The MIT data challenges this directly by changing its shape from a cliff to a slope. The difference between those two geometries is the difference between emergency triage and strategic adaptation, and getting the geometry wrong produces the wrong response.

The secondary distortion is the framing of FOBO as primarily a worker psychology problem. The article’s data tells a different story: 83% of executives say their organizations lack the data infrastructure to fully leverage AI. Only 19% of US establishments have adopted AI at all. The organizational paralysis is larger and better-resourced than the individual paralysis, and it is producing the same outcome (i.e., failure to adapt) while generating considerably less cultural attention. FOBO with a corner office is the same condition as FOBO on the factory floor, and it is at least as consequential.

The deepest distortion is the one EY’s chief innovation officer identifies without naming it as a distortion: the workers most at risk are not the least skilled or the least experienced. They are the most skilled and most experienced workers who built their professional identity around doing things well without AI assistance and are now watching less experienced colleagues outperform them by a factor of ten or twenty using tools they refuse to adopt. FOBO, in this formulation, is not the fear of becoming obsolete. It is the active process of becoming obsolete while believing that excellence at the previous version of the job constitutes a defense against it.

For the executives making dramatic displacement predictions (e.g., Amodei’s 50% figure, Warner’s 35% unemployment projection) the incentive structure varies but the effect is consistent: urgency narratives accelerate adoption, investment, and policy attention. Whether the predictions are accurate or not, they create the pressure that moves institutions. Warner’s observation that AI leaders are “consciously pulling back on their predictions” because of short-term economic disruption concerns is the most honest moment in the article’s framing section because it acknowledges that the predictions are being managed, not just made.

For the MIT researchers, the incentive is empirical accuracy over narrative convenience. The rising tide framing is less alarming than the crashing waves model, less useful for generating headlines, and more useful for calibrating actual institutional response. The study’s careful insistence that “gradualism is not inherently protective” (a slower disruption is still a disruption) is the intellectual honesty that makes the research valuable and the urgency still warranted.

For employers, the incentive gap the article documents is structural: the productivity gains from AI adoption are real and compounding, the costs of non-adoption are real and compounding, and yet fewer than one in five establishments has adopted AI and two-thirds of workers are receiving no training or guidance. The gap between the incentive to adopt and the actual adoption rate reflects not indifference but the organizational friction, data infrastructure deficits, and change management complexity that make enterprise AI adoption genuinely hard — and that the headline productivity numbers systematically underrepresent.

For the resistant senior workers EY’s Depa describes (the ones who have gone from top of their class to bottom of their peer group) the incentive to adopt is real and visible and being actively overridden by professional identity, pride, and the belief that excellence at the old version of the work constitutes a durable competitive advantage. It does not. And the tragedy in Depa’s account is that the workers best positioned to use AI well are the ones most likely to resist it, while the workers least positioned to compensate for AI’s errors are the ones most enthusiastically adopting it.

The immediate consequence is a productivity bifurcation that is already visible inside organizations like EY and will become visible across labor markets as adoption spreads. Workers and companies using AI effectively are compounding their output advantage daily. Workers and companies not using it are falling behind a baseline that is rising without them. The 40-60 minutes per day recaptured by AI users translates to 33-50 hours of recovered productivity per week across a team of 50; a number that does not stay theoretical for long when a competitor is running that math on their payroll.

The structural consequence of the MIT findings is a three-to-five year window for adaptation that is neither infinite nor instantaneous, and that is being consumed, in part, by the organizational paralysis the article documents. The 81% of establishments that have not adopted AI, the two-thirds of workers receiving no training, and the senior leadership cohorts at EY with the lowest adoption rates are all spending time in the window doing something other than moving through it.

The longer-term consequence connects to the developmental sequence we have been tracking across multiple pieces: the brain fry research showed the cognitive ceiling of AI supervision load; the dark factory piece showed the logical endpoint of removing humans from software development; the wage reset piece showed the mechanism by which AI compresses lifetime earnings without producing the visible unemployment that would generate political response. FOBO is the psychological surface manifestation of all of those structural processes operating simultaneously. It is the felt experience of a workforce living through a transition that is too slow to be a crisis and too fast to be managed by the institutions responsible for managing it.

The consequence Depa identifies at the individual level is the most precise: the workers who refuse to adapt “would have to find a different role.” That is corporate language for a career cliff that arrives not through dramatic displacement but through the gradual erosion of relative productivity until the gap between the resister and their AI-augmented peers is too large to justify the compensation differential. FOBO, left untreated, becomes self-fulfilling because the fear produced the paralysis that produced the outcome.

The MIT study’s most important contribution is not its capability data but its timeline calibration. A crashing wave demands emergency response. A rising tide demands strategic adaptation. The responses are different, the institutional requirements are different, and the psychological relationship to the change is different. Getting the geometry right is the prerequisite for getting the response right.

The calibration for workers is the one EY’s data makes concrete: the question is not whether AI will affect your role but whether you are treating the current window as time to adapt or time to wait. The workers thriving inside EY are not the ones who were least vulnerable to AI. They are the ones who decided to treat AI as a tool rather than a verdict and began building the capability to use it before they needed it for survival. The workers struggling are not the least capable. They are the most experienced and the most invested in a version of professional excellence that the tool is in the process of commoditizing.

The calibration for organizations is the infrastructure question the article surfaces and does not resolve: 83% of executives lack the data infrastructure to fully leverage AI, two-thirds of workers are receiving no training, and the adoption rate remains below 20%. These are not individual failures of courage or curiosity. They are systemic failures of organizational investment in the transition that the MIT data shows is already well underway. The rising tide gives you time to move. It does not move the furniture for you.

The calibration for the public conversation is to hold the MIT findings and the EY data simultaneously: the disruption is real, the timeline is longer than the headlines suggest, the window is open, the window is not infinite, and the workers most at risk are are the ones with the most to protect and the strongest incentive to believe that protecting it requires no change at all.

FOBO is rational. The MIT data confirms the direction of the fear while correcting its timeline. The antidote is adaptation (institutional, organizational, and individual) at a pace that the rising tide makes possible and that the current adoption data suggests most organizations have not yet chosen to match.