The hidden talent risk of AI workflow automation

If you’ve been online at all in that past couple of years, you’ll know there’s a lot of excitement about AI productivity gains. But what if we told you that too much AI workflow automation can harm your talent pipeline? 

Deloitte’s most recent Global Human Capital Trends report found 85% of leaders agree it’s critical to build your organization’s and workforce’s ability to adapt continuously, but only 7% say their organization leads in enabling continuous growth. And Fosway’s 2025 HR Realities research found skills availability is no longer the top challenge for organizations, being surpassed by performance and profitability as well as multiple AI-related challenges. 

But with the current pace of tech change, workforce agility and adaptability has never been more important. At Zensai, we believe human oversight will always be essential for getting the best and most trustworthy results from AI. That’s because human workers play a key role by sense-checking AI output and making vital, context-sensitive judgment calls. 

So, what happens when humans in the loop become over-reliant on the very technology they’re supposed to monitor? 

The state of AI workflow automation 

There’s a lot of buzz right now about AI’s ability to automate work, but it’s important to separate hype from reality. According to Deloitte’s The State of AI in the Enterprise report, 82% of companies said they expect at least 10% of their jobs to be fully automated in the next three years. 

According to Deloitte, “Leaders in the qualitative interviews expressed concerns about potential disruption to professional development pipelines as a result of automation. Entry-level jobs involving data entry, reconciliation, and first level customer support at their companies are  being prioritized for automation, but these jobs are often the starting point for longer careers.” 

Although 10% may sound conservative, three years really isn’t long in the grand scheme of things. Gartner goes a step further by predicting that, by 2030, 75% of IT work will be done by humans augmented by AI, and the remaining 25% will be entirely AI-automated. 

The business benefits of AI automation 

Like any form of automation, AI benefits businesses by reducing the number of work hours needed to achieve certain results, reducing costs. While we don’t yet know the full cost-saving potential of AI, we do know it’s a major business priority. 

According to Gartner, 54% of infrastructure and operations (I&O) leaders rated cost optimization as their most important goal for adopting AI. Beyond that, AI workflow automation can benefit businesses in the following ways: 

• Reducing or eliminating time-consuming tasks for worker wellbeing and engagement. 

• Quickly analyzing data and catching easily missed trends for insightful, real-time analytics. 

• Catching human errors to prevent costly rework. 

• Streamlining or automating key HR processes for a better employee experience. 

How limiting AI workflow automation protects your business 

Although AI workflow automation is largely beneficial to businesses, leaning on it too heavily can create issues with your talent pipeline and succession planning. Let’s look at some of the reasons it’s worth holding back on total automation even if it’s feasible. 

The value of humans in the loop 

We touched on this earlier, but a human in the loop is essential for AI workflow automation because real people must be the ones making judgment calls and checking AI output for errors. 

The fact is that AI can never take responsibility for anything it does wrong, such as hallucinating data or recommending an incorrect course of action. But, if issues like these make it past layers of human oversight, the people who missed it absolutely can. 

AI tools also can’t recognize nuance in the same way humans do. For example, an AI performance management tool might flag some employees as “high flight risks” and recommend freezing promotions and pay increases. A human leader, on the other hand, would recognize that this would damage trust and engagement, increasing the likelihood of turnover. 

Businesses already face tech-driven talent gaps 

According to Deloitte’s previous (2025) Global Human Capital Trends report, 66% of managers and executives say their most recent hires aren’t fully prepared for their roles, with the most common issue being lack of experience. 

Similarly, according to Mercer’s Global Talent Trends report for 2026: 

• 54% of executives cite talent scarcity as the number one driver shaping their people strategy. 

• 63% agree they must shift to skills-powered talent practices to prepare for the future. 

• Additionally, 59% of HR leaders say attracting candidates with the right digital skills is their top people challenge. 

• 53% of employees are worried they lack future-ready skills. 

How AI workflow automation can limit critical thinking 

A Microsoft survey of knowledge workers found that 40% of GenAI-assisted tasks involved no critical thinking at all. Higher confidence in AI was strongly associated with reduced cognitive effort, while creation tasks showed the lowest engagement. 

This isn’t to say that all AI use impairs worker thinking. But using it in the wrong ways certainly can. That’s why it’s essential for employers to set well-researched, effective use policies for AI tools to guide employees instead of assuming they’ll find their way. They also need the space and support in their roles to exercise their critical thinking and core job skills on a regular basis. 

Manual skills are essential for judging AI output 

Not everyone can effectively judge the results of AI workflow automation. It takes a good understanding of whatever task a given agent or GenAI tool has been trained to do. For instance, you’d want an experienced coder to check AI-generated web design, or a skilled writer to make sure your blog content hits the mark. 

Without these core skills as a basis, employees will be more likely to defer to AI output without critical analysis. The problem is, if you’re bringing new hires into a work culture that expects them to use AI for these tasks, they’ll miss their chance to build these foundational skills. Right now, organizations are full of employees who remember a time before AI, and who spent a long time honing various job-critical capabilities. 

Fast-forward a few years, however, and they’ll soon be outnumbered by people who never had the same opportunity. That’s why we recommend keeping a certain percentage of AI-automatable tasks manual. This gives new hires an opportunity to cut their teeth, and established employees a way of keeping their skills sharp. 

How AI can support manual skills development 

Despite the potential issues with AI workflow automation, AI can also give you the tools necessary to overcome them. So, to wrap things up, here’s a rundown of the ways you can use GenAI and agentic AI to support employees as they develop the manual skills essential for their roles: 

• Content creation: AI tools can streamline content creation by converting relevant documents into a bespoke training course complete with quizzes and other interactive elements to test knowledge retention. 

• Content curation: AI can deliver personalized learning path recommendations based on role type, strengths, weaknesses, and employee ambitions. That way, employees learn skills that not only matter to them but also have a real impact on your business. 

• Skills coaching: AI agents can coach employees in real time to assist skill development, whether it’s for customer communication, content production, or any other core skill you can think of. 

• Feedback coaching: In the same vein as skills coaching, agentic AI can also coach managers on how to give better feedback that aligns with learning so they can effectively support employee development at every stage of the L&D journey. 

Slightly less efficiency for greater organizational resilience 

It looks like AI workflow automation is here to stay, and this is largely a good thing. Its ability to quickly handle data and streamline busy work creates huge potential productivity gains while allowing employees to focus on what matters most. But, as with most good things, it’s possible to have too much. 

To sustain your talent pipeline in the long run, you need to keep some work manual. That way, employees can gain or refresh their first-hand experiences so they’re better able to oversee the tasks you do automate. 

If you want to learn more, check out our insights on AI productivity, or download our guide, Transformative AI for Learning and Talent Management today and discover how AI can support lasting employee skills growth.