Goldman Sachs warns AI could displace up to 7% of jobs

A new research note from Goldman Sachs analysts Joseph Briggs and Sarah Dong estimated that artificial intelligence could automate about 25% of total work hours over the adoption period, with meaningful implications for workers and employers. Their baseline scenario assumes a 15% AI-driven lift to labor productivity and projects that roughly 6 to 7% of jobs could be displaced during the transition, with a modeled peak increase in the gross unemployment rate of about 0.6 percentage points — roughly equivalent to one million additional unemployed workers at the peak.

The analysis anticipates uneven exposure across occupations. Roles involving office support, routine legal tasks, basic coding and repetitive finance functions are identified as having higher automation risk, while other jobs show lower exposure. The note stresses that displacement is concentrated and sector-specific, rather than uniformly affecting all employees.

Briggs and Dong also place the short-term disruption in historical context, noting that prior waves of automation have ultimately led to significant new job creation even as they eliminated some categories of work. That long-run adjustment, however, does not remove the near-term strain the analysts expect as firms adopt AI tools and reorganize processes.

For workers the findings underscore two clear pressures. Employees in high-exposure roles face a tangible displacement risk and will likely confront altered job descriptions, increased monitoring of productivity metrics and shifting expectations for technical skills. For managers and human resources teams the note signals a need for active workforce planning: targeted reskilling, internal redeployment pathways and updated hiring criteria to reflect AI-augmented workflows.

The report also has policy implications. A temporary uptick in unemployment concentrated in certain occupations could raise demands for expanded retraining programs, more flexible social safety nets and measures to support rapid reentry into employment. Firms may face reputational and retention challenges if layoffs occur without clear plans to transition affected workers.

Goldman Sachs’ modeling offers concrete figures for a debate that has often been abstract: a quarter of work hours could be automated, a modest productivity lift could displace a measurable share of jobs, and peak unemployment could rise by about 0.6 percentage points. For employees and workplace leaders, the immediate task is to map exposure, invest in relevant skills and design transition strategies that limit short-term harm while positioning the organization to benefit from productivity gains. Policymakers and labor groups will be watching whether the transition follows the historical pattern of job creation or produces deeper frictions that require broader intervention.