Research note on AI as a partner
Why the best AI partners are the ones that complement, rather than reflect, your existing capabilities?
When choosing AI tools for your team, what’s your instinct? Most of us gravitate toward systems that feel familiar—AI that works the way we think, processes information like we do, and reinforces our existing strengths. It seems logical: if you’re analytically minded, you’d want AI that enhances your analytical capabilities. If you’re creative, you’d seek AI that amplifies your creative processes.
But what if this intuition is not right completely?
Through a series of experiments involving over 700 managers from the US and UK, we uncovered a fascinating paradox about human-AI collaboration. The research is still in progress, but our findings already reveal clear patterns. When managers worked with AI systems that complemented their skills rather than mirrored them, they didn’t just perform better—they felt more supported, more confident, and more capable of tackling complex decisions.
Our experimental design tested this systematically. Managers first self-assessed their dominant competences—whether interpersonal, technical, or business skills. They were then randomly assigned to work with AI agents specialized in one of these areas. The key insight emerged from comparing their responses: those working with AI that didn’t match their strengths consistently reported greater support in filling their competency gaps.
The findings challenge our fundamental assumptions about productive partnerships. Managers consistently rated complementary AI as more valuable than AI that simply amplified their existing strengths. More intriguingly, this preference wasn’t driven by novelty or curiosity—it was rooted in something more practical: gap-filling.
The managers who benefited most from complementary AI were those who recognized their own limitations. When AI offered capabilities in areas where they felt less confident, it didn’t just provide technical assistance—it provided psychological reassurance. The AI became less of a tool and more of a trusted advisor, filling the spaces where human expertise ran thin.
Perhaps most surprisingly, our experiments revealed that AI matching managers’ dominant skills wasn’t perceived as more familiar or easier to use. This shatters a common assumption in technology adoption: that similarity breeds comfort. Across all our studies, familiarity levels remained consistent regardless of whether the AI matched or complemented managers’ strengths.
What matters isn’t how familiar the AI feels. Still, how useful it proves to be: while managers initially rated complementary AI as less useful on the surface, they simultaneously recognized its value in addressing their weaknesses. This created an overall increase in perceived usefulness—but only when the AI’s recommendations felt relevant to their specific work context.
This creates an interesting tension for organizations implementing AI. The systems that feel most natural at first glance may be the least valuable in the long run. Meanwhile, the AI tools that require a steeper learning curve might deliver the greatest returns—if managers are willing to persist through the initial discomfort.
As AI becomes increasingly central to management and leadership, perhaps the question isn’t “What kind of AI would work best with my strengths?” but rather “What kind of AI would push me to become more than I currently am?”
The mirror shows us who we are. But the best partners—human or artificial—help us become who we could be.
This work is part of the project “People and Algorithms in Organisations: Competences to Work in the Digital Environment” (DIGIT_People and algorithms), funded by the NAWA – Narodowa Agencja Wymiany Akademickiej (Polish National Agency for Academic Exchange).
DIGIT_NAWA AI ArtificialIntelligence Management Leadership HumanAICollaboration ComplementaryAI AIStrategy BusinessStrategy DigitalTransformation FutureOfWork AIResearch NAWA