The first experimental studies which addressed the question concluded that lower performers were getting a bigger productivity boost thanks to AI than higher performers. Some recent study contradicted these results.
The research
In the study where researchers conducted an experiment with 444 professional writers, the conclusion was that initial inequalities in productivity between participant are half-erased by the treatment. In another study which involved consultant from Boston Consulting Group, a similar finding was observed consultants below the average performance threshold increased their productivity by 43% and those above by 17%.
However, in a more recent study which analysed the impact of introducing AI to the 1,018 scientists in the R&D lab of a large U.S. firm, the results were the opposite: the bottom third of scientists see little benefit in productivity but the output of top researchers nearly doubles. Similarly a study including entrepreneurs concluded that the high performing entrepreneurs were benefitting more from the technology (in terms of revenue generation). The reason for these findings expressed in the studies is that the best performers are better equipped to select the interesting suggestions (and not loosing time on the others) and have mode abilities to implement ideas or suggestions.
Reconciling the two seemingly contradictory results, a research has found that different types of AI systems impact people in different ways, depending on how those systems operate. Imagine two types of AI: one that works like an independent expert (autonomous AI) and another that acts more like a helpful assistant (non-autonomous AI). Autonomous AI, which can make decisions and operate on its own, tends to provide the most value to people who are already highly skilled or knowledgeable—like giving a professional chef a state-of-the-art cooking robot. On the other hand, non-autonomous AI, which requires human input and collaboration, benefits those with less experience or expertise the most—similar to how a beginner cook might thrive with step-by-step guidance from a smart recipe assistant.
Implications
These studies have two main implications
- AI investments should be tailored to workforce groups. Different types of AI systems—autonomous for top performers and non-autonomous for lower performers—should be strategically deployed to maximize their impact.
- AI investments should be adapted to strategic objectives. Autonomous AI, which empowers skilled employees to innovate and achieve breakthroughs, can enhance competitive differentiation by driving innovation and maintaining market leadership. Meanwhile, non-autonomous AI can boost the productivity of less experienced employees by providing structured guidance and support.
By understanding these dynamics, managers can strategically deploy AI where it delivers the highest value, ensuring that the organization optimally benefits from AI’s potential to amplify talent across performance levels.
Photo de Kostiantyn Li sur Unsplash
