The Turing Trap is popping right into a Talent Trap: The incorrect AI technique might strand tens of millions of expert employees

In 2022, Prof Eric Brynjolfsson identified and named the “Turing Trap”, the mistake of treating the goal of AI as imitating humans so machines can replace them, rather than building systems that complement and extend human capabilities. 

If the Turing Trap is a mistaken engineering perspective on AI, its damaging counterpart is the Talent Trap, a widely-held but deeply flawed view of workers. In Talent Trap thinking, society has a fixed amount of talent, developed mostly in schools, defined by educational attainment, and companies win by securing that limited pool of talent rather than growing it widely. It says some people are “talent” to be pursued, cultivated, and developed; others are “labor” to be plugged into a well-defined process determined by others, without the autonomy to exercise judgment. 

The Talent Trap worldview treats human workers as fixed realities, not as working learners. So as AI hits benchmarks on more cognitive tasks, those tasks, and eventually jobs, go away, pushing humans into more physically intensive tasks or creating new roles accessible mainly to those with new advanced degrees in emerging fields. It also places the burden of adapting workers to an AI economy mostly upon schools, universities, and formal retraining programs. 

The beliefs that could drive us into an AI Talent Trap are wrong. Human workers’ potential is not fixed, most learning is outside formal education, and job-relevant skills are best developed by working with other people and tools to solve customer, company, or community problems. 

Higher education has a vital role, but employers’ most important talent pipeline is the 10 million U.S. workers who parlay skills gained on the job into higher-paid, in-demand jobs each year, a talent pool five times more than new college graduates. Half these upwardly mobile workers are Skilled Through Alternative Routes (STARs), without college degrees, and they are concentrated in a surprisingly small set of ‘gateway’ jobs. Employers, not schools, shape this talent pipeline. At Opportunity@Work, we’ve spent years studying the skills and career pathways of STARs and other workers, with the goal of helping employers recognize talent they might otherwise overlook.

Gateway jobs are around 75 occupations that account for 62% of the job-to-job changes that lead to higher pay. For example, every year, thousands of customer service representatives (a gateway job) become sales representatives or dispatchers, higher paying jobs that reward the same underlying skills: active listening, reading people quickly, staying calm under pressure, keeping a customer engaged when the stakes are high. Building skills in lower-wage origin jobs, deepening them in gateway jobs, and applying them to reach higher-paying destination jobs is how workers thrive and our talent pool expands.

New research by Opportunity@Work and the Brookings Institution demonstrates that roughly half the opportunities to transition from gateway jobs into higher-paying work are highly exposed to AI’s growing capabilities, which could be used both to automate tasks and augment skills. Talent Trap thinking would treat the skills of the 18.6 million workers in the most AI-exposed gateway jobs as fixed, and focus on automating their roles. If that’s the main use of AI for customer support roles and admin assistants, the blast radius will extend to those still climbing toward those jobs, and to every employer who taps into the skills workers build doing them, like in their future sales force. The consequences of these choices for American industry, workers, and labor market will be much larger than the – also important – pathways for entry level college graduates.  

Instead of walking into an AI Talent Trap, what if business deployment of AI at work aimed to complement the fullest range of human talents to serve customers more deeply, in novel ways?

First, the vast majority of U.S. workers would reskill for redefined roles in the context of their current jobs and in skill-based transitions between roles. Employers’ AI adoption would leverage frontline workers’ tacit knowledge and potential, with administrative assistants leading redesigns of business workflows and medical assistants engaging patients to address unhealthy behaviors. AI-augmenting hard-to-fill destination roles (e.g., in sales) to enable skills-promixate gateway workers (e.g., in customer support) would become a robust talent pipeline strategy. 

Fewer executives would cite AI as a reason or excuse for large layoffs. More companies would take an approach like Walmart, which announced plans to keep headcount steady even as “the skills of every single job will change due to AI,” and saw its stock rise twice as fast as the S&P index in the next 6 months. Regional industry networks would combine technology and talent to create more agile workforces, new capabilities, lines of business, and growing market shares.

Core technology investments would shift. Frontier AI labs currently test and promote their latest models using technical benchmarks that measure their ability to replace humans at tasks, but could instead aim to hit benchmarks assessing the efficacy of humans and AI to accomplish goals and learn together. AI talent tools could leverage data on how workers progress in the absence of paper ceiling barriers to become a corrective to the Talent Trap, not its amplifier. New business models might even emerge, like metered augmentation, in which AI could augment missing skills, help teach those skills on-the-job, then certify a worker’s skill mastery.    

Given its transformative power to deploy the aggregated achievements of human intelligence for virtually any purpose we choose, “AI” should really stand for “Amplified Intention”. The goals we set for AI matter deeply enough to create self-fulfilling prophecies. If we invest trillions of dollars, intending to replace working learners with machine intelligence to solve the same problems we already solve today, we’ll eventually get that outcome. If companies avoid the false assumptions of the AI Talent Trap and invest instead in AI technology to augment and develop the fullest range of human talent and contextual knowledge to tackle unsolved problems, it could lead to a cognitive industrial revolution and workforce transformation that we can only begin to imagine. 

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.