Artificial intelligence is reshaping workplaces faster than previously projected and now has the potential to affect 93 per cent of jobs, according to new research from Cognizant.
The consulting firm’s “New Work, New World 2026” report found AI could handle $4.5 trillion worth of work tasks across the U.S. economy today. However, the research also emphasized that human involvement remains essential to capturing the full value of AI investments.
“We’re seeing significant capital flow into AI, and the rapid adoption of these technologies is reshaping the workplace,” said Ravi Kumar S, CEO of Cognizant. “Our research shows that enterprises could unleash $4.5 trillion in labor productivity today.”
Exposure scores rising faster than forecast
The report analyzed 18,000 tasks and 1,000 jobs in the O*NET labor database to assess how work could be assisted or automated by AI. It found the average exposure score across jobs is 39 per cent today, which is 30 per cent higher than what Cognizant forecast for 2032 in its original 2024 report.
Exposure scores measure the degree to which a job can be assisted or automated by AI. These scores are now increasing by nine per cent annually, compared with two per cent in the original research.
Some professions saw dramatic changes in exposure scores. Legal roles jumped from nine per cent to 63 per cent, education from 11 per cent to 49 per cent, healthcare practitioners from 10 per cent to 39 per cent, and CEO positions from 25 per cent to 60 per cent.
Manual work and knowledge work both affected
The research revealed AI’s impact is evolving across different types of work. Roles in computer and mathematics no longer top the exposure scores, indicating AI advancements may have reached limits in some knowledge work areas, according to the report.
Meanwhile, manual labor jobs are being affected faster than previously understood. Transportation roles saw exposure scores rise from six per cent to 25 per cent, while construction jumped from four per cent to 12 per cent.
Non-automatable tasks declining
The percentage of tasks that cannot be automated by AI dropped from 57 per cent on average in 2023 to 32 per cent today, according to the report.
However, AI remains unable to automate upwards of 40 per cent of management, business and financial operations, and administrative tasks. This indicates human expertise and collaboration remain necessary, the research found.
Human skilling essential to value capture
Kumar said businesses must integrate contextual intelligence, build flexible systems, and prioritize human learning alongside technological advancements to unlock AI’s true value.
“Human skilling becomes the bridge through which today’s AI spending translates into tomorrow’s tangible results,” he said. “AI’s promise is realized when we empower our workforce with digital fluency, adaptability and continuous learning, while ensuring AI solutions are deeply contextualized to unique business challenges.”
The report emphasized that much of AI’s value potential remains untapped without human intelligence and involvement. The technology’s potential lies in freeing workers for higher-order thinking, creative problem-solving, and developing novel ideas that can transform business operations, according to Cognizant.
Cognizant, based in Teaneck, N.J., released the report Jan. 15. The company operates a Synapse program that aims to reach two million individuals globally by 2030 to prepare workers for future jobs.
