Enterprises Accelerate AI Autonomy, Yet Trust Deficit Remains to Be Bridged
A Kyndryl survey shows that only one in four business and technology leaders fully trust their AI systems, but more than four in five respondents expect autonomous AI agents to make decisions with substantial business impact within the next year. AI is now widely embedded in core processes, and enterprises are adapting by redesigning roles and upskilling.

At a Glance
- Despite low trust in AI systems, businesses are accelerating their autonomy. According to a survey released by Kyndryl on Thursday, only a quarter of business and technology leaders say they fully trust their AI systems. The service provider surveyed 1,100 business and technology leaders across eight countries between March and April.
- Even so, the technology is already deeply embedded in enterprise workflows. The survey found that more than four-fifths of technology leaders expect autonomous AI agents to make decisions with substantive business impact within the next year; two-thirds of respondents said they have already given AI autonomous read-write access to their core systems.
- Kyndryl found that the growing reliance on AI highlights the need to adjust talent strategies. Kyndryl Chief Human Resources Officer Mark Paulek said in a statement that leaders should align employees' skills and decision-making with changes in how work is done. "When people understand their role in the system, trust and performance improve together," he said.
Deep Dive
As agent autonomy increases, organizational governance and accountability mechanisms—as well as how to train employees to use AI—will become key factors in successful AI adoption.
Governance approaches vary by organization and team, but a May report from Gartner noted that adopting a proportional governance approach, setting different levels of strategic authority for different agents, can prevent organizations from overextending autonomy.
The Kyndryl survey shows that AI is being more broadly embedded into core business processes, with 57% of technology leaders saying AI is fully integrated, up from 35% last year. Many leaders say AI has not yet significantly changed their daily work, but they are rethinking how humans and AI collaborate.
More than six in ten respondents said they have redesigned roles to accommodate AI, and nearly a quarter said they are creating entirely new roles focused on AI management. The vast majority (95%) of respondents believe that upskilling existing employees is preferable to external hiring.
Kyndryl Chief Information Officer Kim Basile said in a statement that organizations that guide employees through AI transformation via upskilling and role redesign achieve more positive outcomes than those that have not aligned their workforce around AI.
"This is a pivotal moment for global enterprises as they race to adopt AI, redesign workflows, and pursue innovation," Basile said. "However, they are discovering that their greatest asset—their people—requires more attention."