Businesses risk creating an “AI Talent Trap” by using artificial intelligence to replace workers rather than improve their skills and expand their opportunities, an analysis drawing on research by Opportunity@Work and the Brookings Institution has warned.
The argument challenges the assumption that talent is a fixed pool, largely defined by educational qualifications, which companies must compete to secure. It says workers should instead be viewed as “working learners” whose abilities develop through experience, collaboration and the use of new tools.
The warning follows the identification of a related problem known as the “Turing Trap”. In 2022, Professor Eric Brynjolfsson used the term to describe the mistake of designing AI to imitate humans so that machines can replace them, rather than developing systems that complement human capabilities.
According to the analysis, treating human potential as fixed could result in cognitive tasks being automated and workers being pushed into more physically intensive roles. It could also make career progression dependent on advanced qualifications in newly emerging fields, while placing most responsibility for adaptation on schools, universities and formal retraining schemes.
Opportunity@Work estimates that 10 million US workers each year use skills gained on the job to move into higher-paid roles. That is five times the number of new college graduates, while half of those advancing workers do not have university degrees and are classified as Skilled Through Alternative Routes, or STARs.
The analysis says employers, rather than education providers alone, play the central role in developing this talent pipeline. It identifies about 75 “gateway” occupations that account for 62 per cent of job-to-job moves leading to higher pay.
Customer service is given as an example. Workers in those roles can progress into sales or dispatch positions, using transferable abilities such as active listening, quickly assessing people, remaining calm under pressure and maintaining customer engagement during difficult situations.
AI exposure threatens progression routes
Research by Opportunity@Work and Brookings found that roughly half of the opportunities to move from gateway jobs into higher-paid work are highly exposed to AI’s growing capabilities. The technology could automate parts of those roles, but could also be used to strengthen workers’ skills.
About 18.6 million workers are in the most AI-exposed gateway jobs, the analysis says. If customer support and administrative roles are primarily automated, it argues, the impact could extend beyond the jobs themselves to workers trying to progress through them and employers relying on the skills developed there.
An alternative approach would involve employees in redesigning their roles and using their practical knowledge to shape how AI is introduced. Administrative assistants could help redesign business processes, while medical assistants could use technology to support conversations with patients about unhealthy behaviours.
AI could also be used to strengthen hard-to-fill, higher-paid positions such as sales, creating a pathway for workers with closely related experience in customer support. Under this model, reskilling would take place largely within existing jobs and through transitions between roles.
The analysis points to Walmart’s plan to keep its headcount steady while acknowledging that “the skills of every single job will change due to AI” as an example of this approach. It also suggests that regional industry networks could combine technology and workforce development to create new capabilities, business lines and market opportunities.
Such a shift would require changes to the way AI systems are assessed. Instead of relying mainly on technical benchmarks measuring whether models can perform tasks in place of people, developers could measure how effectively humans and AI work together to achieve goals and learn.
The analysis also proposes that AI talent tools could track how workers progress without being restricted by degree requirements. One possible business model would allow technology to provide missing skills, teach them on the job and certify that a worker had mastered them.
Its central contention is that the outcome of AI in the workplace will depend on the objectives set by businesses and technology developers. Investment aimed at replacing workers could reinforce existing roles and assumptions, while investment focused on augmenting a wider range of human abilities could produce a broader transformation of work.
