Follow ZDNET: Add us as a preferred source on Google.
Key Insights from ZDNET
- Even with widespread AI integration, employee confidence in the technology has declined.
- A potential reason is the insufficient training provided.
- Organizations are seeking methods to alleviate worker frustration.
For every task where AI enhances efficiency and saves time, there are numerous instances that leave Tabby Farrar and her team feeling that the technology falls short.
As the head of search at Candour, a UK-based SEO and web design agency, Farrar observes a vibrant discussion regarding AI in the digital marketing sphere. While her team is enthusiastic about the potential of AI for accelerating workflows and reclaiming time spent on mundane tasks, the actual implementation can often yield frustrating results.
Also: 95% of business applications of AI have failed. Here’s why
AI has the capability to generate product lifestyle images for clients lacking any, yet it struggles with critical tasks like creating executive data summaries, frequently missing essential details. Fine-tuning prompts to organize datasets can be so time-consuming that it renders manual efforts more appealing.
“As a manager, I’m aiming to get my team more engaged with AI, recognizing its future relevance across industries,” Farrar stated. However, she added, “Many are expressing frustrations about losing hours trying to make the technology functional.”
Farrar’s experience reflects a broader sentiment as many navigate the disparity between AI’s promises and its actual performance, resulting in growing skepticism.
Workplace Anxiety: A Real Concern
A recent study by workforce solutions provider ManpowerGroup revealed a 18% decline in worker confidence in AI for the first time in three years, alongside a 13% increase in adoption year over year. This disconnect may suggest that the initial excitement surrounding AI is waning, highlighting a crucial moment for companies in their AI implementation strategies.
“An intimidated workforce can’t operate at peak productivity. Such anxiety will create significant challenges,” remarked Mara Stefan, VP of Global Insights at ManpowerGroup.
Other research corroborates this perspective. An EY report from November indicated that while nine out of ten employees utilize AI in their roles, just 28% of organizations can translate this into meaningful outcomes.
“Our findings reveal a fundamental issue: employees might be recapturing a few hours here and there, but the technology isn’t transforming work processes or driving business performance,” the report stated.
For some leaders, tackling this decline in confidence has become a part-time responsibility.
Also: 5 ways you can stop testing AI and start scaling it responsibly in 2026
Randall Tinfow, CEO of REACHUM, an AI-centered learning platform in Scranton, PA, dedicates approximately 20 hours each week out of his 70-hour workload to evaluating AI tools and partners to ensure his team is not overwhelmed by ineffective options.
Although platforms like Claude Code have significantly sped up software development at REACHUM, not all tools meet expectations. Tinfow notes a disparity between how some AI solutions are advertised and their actual effectiveness.
Even within an AI-focused organization, Tinfow’s team has faced challenges, such as ineffective text generation for images—tasks that some AI tools simply cannot handle adequately.
“There’s so much misleading information, and I want to prevent my team from falling for it, so I review any new tool first, determining its practicality before sharing it with them,” Tinfow explained.
Restoring Confidence in AI
This disconnect between expectations and reality may be a significant factor behind the declining confidence, according to Kristin Ginn, founder of trnsfrmAItn, an organization dedicated to assisting businesses in AI adoption while considering the human workforce.
While marketing presentations showcase the ease of use, business leaders must ensure employees understand that trial and error will be an essential part of the process.
A psychological factor also contributes to the issue. ManpowerGroup’s study indicated that 89% of respondents feel confident in their existing roles. Many have performed their jobs the same way for an extended period.
Also: 5 ways rules and regulations can help guide your AI innovation
“When exploring how to incorporate AI into familiar tasks, employees must exert significantly more mental effort to adapt to this new approach,” Ginn pointed out. “This disruption of routine changes their confidence in their work, emphasizing human tendencies to resist change.”
Stefan highlighted the importance of providing adequate training to sustain confidence levels. Over half of respondents (56%) indicated they have not received recent training, while 57% reported a lack of mentorship opportunities.
“Organizations that effectively address these issues, making employees feel confident in using technology through proper training and contextual support, will greatly benefit,” Stefan remarked.
Identifying Opportunities
Back at Candour, Farrar noted various strategies to balance the pursuit of innovation with daily technological challenges.
The agency allocates additional time for training, frames experiments as “test and learn” to alleviate stress, and appoints an “AI champion” to monitor technological advancements. The CMO has conducted training sessions, and Farrar regularly checks in with her team, admitting her own frustrations at times.
Also: Transform AI chaos into a career opportunity by preparing for these 4 scenarios
Some of these initiatives have yielded positive results, such as the development of a Gemini Gem, trained to adhere to brand and tone-of-voice guidelines to produce client-approved quotes for media use. Candour’s innovation lead is actively creating tools tailored to their specific needs using APIs from companies like OpenAI. Farrar illustrated how their perspective on AI imagery improved rapidly following the release of Google’s Nano Banana.
Nevertheless, there remains a considerable journey ahead.
“If I am going to delegate some responsibilities to these tools,” Farrar concluded, “I need to trust that they can perform at least as well as I do.”
