Employees, managers, and organizations today are facing a growing disconnect: work is changing due to widespread use of AI, but not in a coordinated or intentional way. The problem certainly isn’t a lack of ambition around AI. Rather, it’s a clearer way to incorporate AI into the work itself. The Future of Work Research Lab at SAP has been conducting research to understand what makes work fundamentally suited to humans or AI, how the use of AI can bolster human thriving, and a model to apply that understanding through work redesign.

How jobs are changing today

In a recent study of 4,016 global employees and managers, we uncovered the ways in which work has changed over the past year. Findings show that certain human-oriented tasks are on the rise: leading, managing, and supervising (46% reported an increase) and problem-solving (45%). Nearly half (45%) are spending more time learning new information, skills, or tools, likely due to increased AI adoption. The tasks that are decreasing in frequency are consistent with AI’s strengths today: routine administrative work (19% reported a decrease), creative work (17%), and analyzing data and creating reports (17%). While these changes are likely a result of a variety of factors, 88% of the sample report that the changes were due, at least in part, to their use of AI at work.

Most interestingly, however, is who instituted these changes—if anyone at all. In just over half of cases (51%), jobs changed with no direction or visibility: employees decided to make these changes themselves without guidance from their organization or manager (32%), or they report the changes just happened to them naturally over time (18%).

Work isn’t being intentionally redesigned—it’s drifting. When work evolves without intention, things start to break: organizations and managers lack visibility into what employees are doing, and if they are doing it safely with the right tools. Employees who redefine their own roles without guardrails might become overworked, burned out, or simply confused about what they’re supposed to achieve. And employees who don’t redefine their roles might get left behind.

Redesigning jobs with intention

So, what are the ways that organizations can intentionally redesign jobs to incorporate AI? In research done for our report The Road Ahead: Predictions and Possibilities for the Future of Work, we uncovered two approaches: “AI maximalist” and “symbiotic strategist.”

The mantra of the AI maximalist approach is if AI can do it, AI should do it. Redesigned roles are structured around maximizing automation, with the tasks that remain delegated to humans. This approach is likely to achieve short-term efficiency and productivity gains, improving the speed and scale of output. But framing human contribution as merely “what’s left over” creates mistrust and demotivation that, over time, will negate those benefits. When work is reduced to what can be automated, value is left on the table and people are left behind. What looks like efficiency in the short term becomes erosion over time.

The symbiotic strategist approach is guided by both AI and human potential. Roles are designed such that humans and AI can combine strengths to generate outcomes greater than what each could achieve alone, maximizing both value from AI and the factors that make work meaningful and motivating for humans. The result is workers that are more engaged, less burned out, and better equipped to contribute to the organization’s mission in novel ways. In this model, AI doesn’t replace human value; it transforms and amplifies it.

A strengths-based approach to task allocation

My colleague, Julie Bartholic, VP of Customer Research, and I have been studying how to tangibly redesign roles in this mutually beneficial way. We have developed a six-step model of job redesign through subject matter expert interviews and working sessions and we are currently partnering with organizations to test this model on real-world jobs. Through this work, we’ve found that the most important, complex, and dynamic step in this process is reviewing and reallocating individual tasks to humans, AI, or a combination of both.

In a symbiotic strategist approach to task review and reallocation, instead of asking, “Can AI do this?” we ask a better question: “Who should do this work – humans, AI, both working together – given the strengths and risks of humans and AI?” That single shift in question is the difference between redesigning work and simply redistributing it. There isn’t a single, stable answer. The strengths and risks of AI are constantly changing, and cultural boundaries around what AI should or should not do vary widely. The solution is to assess tasks on a set of dimensions where humans and AI fundamentally differ:

1. Judgment-based versus rules-based. Whereas humans excel on tasks with high nuance, unpredictability, and ambiguity, AI excels on tasks that are repeatable and predictable with few edge cases.
2. Data quality. AI performs best when grounding data is structured, labeled, and stable; otherwise, humans are better equipped to make sense of information that is sparse, conflicting, or lacks context.
3. Emotional sensitivity. People largely prefer humans to complete tasks that are emotionally laden or require building trust.
4. Risk. Humans are needed in the loop for high-risk tasks that require accountability and explainability and could result in legal exposure, safety issues, or reputational harm, whereas AI can autonomously handle tasks where explainability is less important and negative outcomes are minor inconveniences or reversible.
5. Scale of output. AI has the advantage over humans for high-volume and repetitive tasks.
6. Speed. Humans outperform AI when time is allowed or needed for reflection, challenge, and escalation; AI excels when decisions are needed in real-time.

Together, these dimensions turn guesswork into design. As technology, strategy, and culture evolve, so too can the thresholds and trade-offs organizations use related to these dimensions. While this list still includes productivity (#5) and efficiency (#6), it treats them as just part of a broader set of strengths dimensions. This reframes the conversation from where we can add efficiencies, to where we can maximize strengths.

Recall the task that is increasing in frequency the most—leading, managing, and supervising. If we look at this task through our six dimensions, we can understand the reason for this shift. For example, managing different people in different circumstances requires different responses that can only be derived from human judgment, not a standard set of rules. Additionally, managing through complex changes or having tough performance conversations require high levels of emotional sensitivity, best delivered by a human. Conversely, the task that is decreasing the most—routine administrative work—falls under AI’s strengths: completing large-scale outputs at high speed according to a standard set of rules.

Conclusion

AI will continue to change what is possible at work, but it won’t decide what work should look like. That is a design choice that belongs to employees, managers, and organizations. And the businesses that treat it that way—grounding redesign decisions in the fundamental strengths of humans and what improves their experience—won’t just adapt to the future of work. They’ll define it.

Watch this space for the full research report, coming later this year where we explore these implications in greater depth and potential directions for HR technology.