Workers are warming to agentic AI, but concerns over accuracy linger

Agentic AI concept image showing human brain split in two halves, with one side digitized on a grid pattern and the other illuminated in yellow with brainwaves emitting.
(Image credit: Getty Images)

Most workers in the US and UK are using AI agents at work, according to new research, despite the fact that a third don't trust their reliability or quality.

Nearly six-in-ten are using AI agents on a daily basis, and told YouGov researchers commissioned by Pegasystems that they're seeing useful results.

Four-in-ten highlighted the automation of tedious tasks as the main benefit of agentic AI, followed by reduced time spent searching for job-related information and quick meeting summarization, both cited by around a third.

However, workers have big reservations that prevent them from fully trusting agentic AI. A third said they were worried about the quality of work produced by agents, with a similar number pointing to a lack of human intuition and emotional intelligence.

Similarly, three-in-ten said they don’t trust the accuracy of AI-generated responses.

"Both organizational leaders and their employees recognize the significant benefits of agentic AI," said Pegasystems CTO Don Schuerman.

"However, this new research underscores that many still have reservations, and it’s up to enterprise leaders to strategically and thoughtfully incorporate the technology to help ensure adoption."

As a result of these concerns, four-in-ten are uncomfortable submitting AI-generated work while 34% worry that AI-produced work isn’t as good as their own.

Despite this, nearly half still believe AI will positively impact their jobs over the next five years, with only 13% anticipating negative effects.

Nearly half (42%) of respondents identified enhanced accuracy and reliability as the top priority for improvement in agentic AI tools, followed by better training on how to use these tools (39%), and increased transparency in AI decision-making processes, cited by a third.

Making agentic AI work

To make agentic AI work and achieve mass adoption, organizations will need to put in some effort, according to Pega. This means developing an orchestrated approach, with AI designing, automating, and optimizing workflows and customer journeys, but only alongside humans.

"Organizations must meet employees where they are by integrating AI agents with actual workflows so they’re not just doing any work, but doing the right work," said Schuerman.

"This, combined with proper governance, transparency, and educational opportunities, will be vital for maximizing productivity, increasing comfort with the technology, and ultimately achieving widespread adoption for a true return on investment."

In another report released this week, digital operations management company PagerDuty found that more than half of CIOs and CTOs see agentic AI as core to their future IT operations.

RELATED WHITEPAPER

Nearly four-in-ten said they had witnessed operational efficiency gains from the technology, with 36% saying it had improved customer experiences and 38% that it was giving them better insights from data.

"AI has rapidly grown to become the backbone of digital operations, helping reduce toil, improve resilience and drive tremendous efficiency," said Katherine Calvert, chief marketing officer at PagerDuty.

Emma Woollacott

Emma Woollacott is a freelance journalist writing for publications including the BBC, Private Eye, Forbes, Raconteur and specialist technology titles.