Human Resources
Not All Friction Is Bad Friction in AI Automation
AI is changing how work is done every day—and making human judgment and expertise more important than ever.
Sara Braun
Editorial Strategist, HR
Workday
Two things can be true at once. AI agents enable unprecedented automation, but human judgement remains essential.
Workday’s most recent AI@Work Pulse found that 57% of employees say they can’t rely on AI outputs without checking them closely. One-third say that at least half their time “saved” using AI is later consumed by correcting or reworking low-quality outputs. The knee-jerk reaction is to try to remove this back and forth. But is all friction bad friction?
As AI becomes part of the business operating standard, leaders need a clear way to distinguish friction worth eliminating from effort that drives meaningful results. A more useful view of AI-driven efficiency looks beyond hours saved and asks whether AI is helping employees do better work.
The next opportunity is to shift from simply getting employees to use AI more to helping them use it well.
Report
AI Agents Are Here—But Don’t Call Them Boss
When Friction Adds Value
For years, automation strategies have primarily focused on eliminating repetitive tasks and creating more time for higher-order work. AI dramatically expands the ability to do this, making many kinds of work faster and easier. But applying the same automation logic to every task can remove more than inefficiency.
Manual data entry, scheduling, standard reporting—this is the kind of routine work AI can and should help reduce to free workforce capacity. But other tasks require more nuance and care. High-stakes work that calls for interpreting complex insights or weighing decision tradeoffs might happen slower precisely because it requires human judgment.
Recent labor-market data suggests that judgment may become more valuable as AI takes on more routine work. PwC’s 2026 Global AI Jobs Barometer found that jobs “professionalized” by AI—where technology handles more basic work while human expertise becomes more important—are growing twice as fast as jobs where AI makes the work easier for non-experts.
In other words, AI is not just reducing the need for human effort; in some roles, it is raising the value of the expertise that remains.
Automate Tasks, Not Learning
Employees build knowledge and expertise every day by researching problems, troubleshooting, making hard decisions, and seeing where their reasoning needs improvement. Early on in a career, those experiences create the foundation for handling more complex and higher-stakes work.
When AI absorbs too much foundational work, organizations also remove some of the experiences employees need to develop professionally.
With AI changing internal learning paths, development can no longer be treated as a byproduct of routine work. Organizations need to be intentional in ensuring AI does not take over the parts of the job that foster knowledge development, and that employees still have a clear view into the why and how behind workflows and decisions. Otherwise, they will never have the human judgement and expertise needed to succeed.
For example: An analyst generating reports should understand AI’s data sources and why they’re being used. A marketer using generative AI should still maintain full knowledge of brand standards and values. An operations manager should still know exactly how things are executed on the floor, even as AI contributes to process optimization.
It’s up to leaders to strike a balance between making employees more productive and ensuring they never lose the human context that keeps them accountable for their work.
AI adoption is increasingly becoming a work design challenge.
A Framework for AI Work Design
For HR leaders, AI adoption is increasingly becoming a work design challenge. Decisions about what to automate shape how roles are structured, how employees can build capabilities, and where accountability lies in the org chart as AI takes on more work.
A practical way to make those decisions is to sort work into three main categories: automate, augment, or preserve.
Category 1: Automate
Full automation is the right fit when human involvement does not meaningfully improve the outcome of the work. These are tasks where reducing manual effort can improve efficiency without weakening quality, ownership, or human development.
This may include:
Routine scheduling and coordination
Data entry and system updates
Document retrieval
Standard administrative workflows
Repetitive reporting or information gathering
It’s important that the automation only goes this far, and decisions aren’t made without further context. A recurring report may be easy to automate, while interpreting the findings and deciding what to do next still requires judgment.
Category 2: Augment
Augmentation is the right fit when AI can strengthen the work, but a person still needs to guide the outcome. In these cases, employees can use AI to move faster, see patterns, or develop a stronger starting point while still applying their own judgment to the final decision.
This may include:
Analyzing workforce or business information
Preparing for a difficult conversation
Drafting an initial communication
Surfacing patterns or recommendations
Modeling possible scenarios
In these cases, the employee should still be able to genuinely assess AI outputs, recognize when they don’t meet requirements, and make decisions about next steps. If employees no longer understand the work well enough to challenge outputs, their role becomes a form of rubber-stamping rather than meaningful review.
Category 3: Preserve
Some work should remain primarily human because the person directly shapes the outcome. These are moments where quality depends on human trust, accountability, lived context, or the relationship between the people involved.
This may include:
Coaching and career development
Conflict resolution
Sensitive employee conversations
Relationship-building
Complex or high-stakes decisions
Work that helps employees develop expertise through practice
Preserving the human role doesn’t mean keeping AI out entirely. But in this category, the person remains the primary actor. AI can help with preparation, context, or administrative follow-up, but it should not become the main driver of the work.
The goal is not simply broader AI use—it’s better work.
A More Intentional Standard for AI
The next phase of AI adoption will require a more thoughtful approach to efficiency. As AI becomes capable of handling more tasks, organizations will need to make clearer decisions about where it should take over, where it should support employees, and where human involvement still carries the value.
That standard matters because the goal is not simply broader AI use—it is better work. When leaders evaluate automation through that lens, they can reduce low-value effort without weakening the judgment, accountability, and expertise the business still depends on.
Eighty‑seven percent of employees say AI boosts their confidence in decisions when they trust the underlying system. Discover how to build that trusted operational core in this Workday report.
Report
AI Agents Are Here—But Don’t Call Them Boss
A gap between effort and impact is the real ROI problem. Discover the solution.