Cognizant finds four in five workers want more AI use than their jobs permit

Employers face an AI adoption problem that enthusiasm alone cannot solve. Four in five employees want to use AI more than their jobs permit, according to Cognizant’s workplace study. Yet dissatisfaction also runs through its most active users, pointing to a divide between wanting AI at work and having tools that fit the job.

Cognizant published the findings in an October 9 release for “New Frontiers, New Frictions: Mapping the DNA of AI Adoption.” It calls the divide between employees’ perceived opportunity and permitted use the “Activation Gap.” Its proposed remedy combines tools built around specific work, better integration and training tailored to different employees—not simply a wider invitation to use AI.

Heavy use does not mean good tool fit

Across the study’s employee groups, 72%–82% said their current AI tools did not meet their needs. That range includes people who use AI heavily and people less inclined to adopt it. The most satisfied group, labeled Precisionists, consisted of highly active users. Even there, only 28% reported satisfaction with their tools.

The complaints were more specific among two other engaged groups. Cognizant says 33% of Accelerators, its heaviest users, and Architects, its fastest adopters, identified missing context and insufficient customization as the top factor limiting use. Those findings distinguish access to an AI system from access to one equipped for a particular team’s work.

CEO Ravi Kumar S argues that employers should shape tools around a team’s rules, workflows and ways of carrying out tasks. That recommendation also aligns with Cognizant’s business: the company describes itself as a technology services provider building AI solutions for clients. Its preferred response to the findings is therefore a company recommendation, not a demonstrated result of an intervention.

Training succeeds unevenly—and does not fix the tools

The training results show a different divide. Architects reported both higher completion and greater effectiveness than Holdouts, the group least inclined to use AI. But Architects’ relatively positive training experience did not keep them within approved systems: 87% said they turned to unsanctioned tools when authorized options fell short.

Cognizant treats these as separate problems. Training is not reaching every group equally, while approved tools still disappoint employees whose training is effective and well attended. More instruction, on its own, is not the solution the company draws from these findings. It recommends improving the tools alongside adapting training to employees’ different needs.

Different barriers call for different responses

The study says disposition—how someone relates to technology, risk and work—is roughly six times more predictive of AI behavior than their place in the organizational hierarchy. Cognizant identifies two groups, together accounting for 72% of workers, as its biggest opportunity for targeted adoption efforts:

• Navigators are willing to use AI but face organizational barriers.
• Stalwarts need a clearer reason and business case before adopting AI.

Those categories separate an employee who wants permission or better support from one who still needs convincing. The release does not provide the study’s sample size, geographic coverage or survey dates, limiting how confidently readers can apply its percentages to a particular workforce. The findings describe employees’ reported needs, rather than measured productivity gains.

Cognizant places a large economic estimate beside that workplace diagnosis: a uniform 5% increase in global labor productivity could add roughly $6 trillion to world GDP. That is a conditional estimate, not a gain measured in this study or a forecast that better workplace tools will deliver that amount. The employee findings identify friction; the GDP calculation describes the potential scale of productivity improvement.