AI raises employee performance to team level
Quality of results increases, required working time decreases, satisfaction increases
A recent field study (March 2025) investigated how AI support affects employee performance and collaboration. The results are highly relevant for managers: Artificial intelligence (AI) can increase individual performance to such an extent that individual employees with AI can approach the effectiveness of teams. At the same time, it was shown that AI can replicate certain quality aspects of teamwork and even have positive effects on employee motivation and collaboration. The key findings of this study are presented below in a clear and comprehensive manner - from performance improvements and qualitative changes in the solutions to social aspects and named limitations.
The study "Teamwork versus Artificial Intelligence? A Field Experiment in a Product Innovation Task" (2025) was conducted by a team of researchers from Harvard Business School together with Procter & Gamble (P&G). A total of 776 experienced P&G employees from various departments (Research & Development and Marketing/Sales) took part.
During one-day workshops, they worked on real product innovation tasks from their business area - e.g. developing new product ideas, packaging or sales strategies. The participants were allocated by lot to four groups: alone or in teams of two (one R&D and one marketing specialist) and with AI support (access to ChatGPT-4) or without AI.
This created comparative conditions: from individuals without AI to teams with AI. Everyone in the AI group received a brief introduction and sample prompts in advance to enable them to use the AI tool effectively.
The quality of the participants' solution contributions was assessed by technical experts. In addition, the researchers recorded the time required to complete the task and the emotional reactions of the participants (e.g. satisfaction, stress) via a survey. This experimental design made it possible to analyse the effects of AI support and teamwork in isolation and in combination.
Quality of results
The evaluation of performance quality showed impressively that AI support significantly improves the results of individual employees.
Without AI, teams achieved better solutions than individuals, as expected.
With AI, however, individuals were able to achieve a similarly high quality as a classic team of two without AI. In other words, an individual with AI performed as well as two individuals without AI, virtually cancelling out the traditional benefits of teamwork in terms of performance.
The following findings can be summarised in terms of performance:
- Team vs. individual (without AI): teams achieved significantly higher quality scores than individuals without AI. This result confirms the usual advantage of teamwork, as two people can combine their knowledge and ideas.
- Individual with AI vs. team without AI: With AI support, individual employees achieved a similarly high solution quality as teams of two without AI. The increase in performance through AI was very clear, so that one employee with AI was on a par with a conventional team. AI was therefore able to partially replace the missing human team partner.
- Team with AI: The best average quality was achieved by teams that used AI. Interestingly, this advantage over individuals with AI was not statistically significant - in other words, on average, teams with AI were not significantly better than individuals with AI. This emphasises how powerful AI can make individuals. However, teams with AI were particularly successful in producing top solutions. The very best results (top 10% of quality ratings) occurred disproportionately often in the group Team with AI. The combination of several human minds plus AI therefore still appears to be favourable for outstanding solutions.
Productivity
In addition to the increase in quality, there was also an increase in efficiency: AI-supported groups worked faster. Both individuals and teams with AI needed around 12-16% less time for their tasks than their colleagues without AI. Despite the shorter processing time, the solutions developed by the AI group were both more comprehensive and more detailed than those of the control group without AI. The AI therefore not only helped to increase quality, but also to achieve results more quickly - an important productivity aspect for companies.
Process automation
at a fixed price!
Contact us now.
AIMAX Business Solutions combines excellent solutions with first-class service. Your added value is our goal. Unique AI systems allow us to act independently of the application. With process automation and digital assistance, we unlock new potential in your company.
A key qualitative finding of the study is that AI support leads to more balanced, holistic solution proposals and breaks down specialised silos. As expected, without AI, the participants worked very much in their respective specialist mindsets: R&D specialists predominantly proposed technically orientated solutions, while marketing/sales experts tended to contribute market- and customer-oriented ideas. In cross-functional teams without AI, these different perspectives could be combined, which led to more balanced team solutions. However, as soon as AI came into play, these differences almost completely disappeared.
- Without AI support: clear silo effects - the solution ideas varied greatly depending on the specialist background. Individual R&D employees tended to provide technical concepts, while commercial specialists primarily made market-orientated suggestions. Only in the classic team without AI were these two perspectives united through collaboration in order to create balanced solutions.
- With AI support: Regardless of background, all participants developed balanced solutions that included both technical and commercial aspects. AI appeared to provide missing domain knowledge or food for thought from the other specialism, so that R&D and marketing professionals with AI both took a holistic perspective. The technical difference between the contributions of developers and marketers was almost completely blurred by AI - an indication that Generative AI helps to break down knowledge silos.
This effect was particularly pronounced among less experienced employees. Without AI, this group achieved significantly weaker results, even in teams, as they lacked experience and specialised knowledge. With AI support, on the other hand, even less experienced employees were able to develop solutions that were on a par with teams of highly experienced specialists. Here, AI acted as a knowledge broker and "leveller" by compensating for a lack of expertise. Overall, AI allowed participants to think and create beyond their usual specialised boundaries. For companies, this means that AI tools can help employees to act more broadly and tackle more complex problems without being strictly limited to their own silo thinking.
Surprisingly, the study showed that AI support not only has a positive impact on factual performance, but also on employees' work motivation and mood. People who worked with AI reported significantly more positive emotions such as excitement, energy and enthusiasm and fewer negative emotions such as anxiety or frustration. This pattern occurred in comparison to participants without AI and was so strong that individuals with AI had a similar positive emotional experience as human teams of two. In other words, working with the AI tool felt more motivating for many participants than working alone without support - almost as if they had a helpful colleague by their side.
These findings are remarkable insofar as technological upheavals (especially AI) are often accompanied by concerns about stress, excessive demands or declining satisfaction. In the experimental context of P&G, however, the opposite was shown: the AI acted as a kind of "social factor" that lifted the mood. The dialogue-based interaction with the AI tool apparently made employees feel supported and less isolated. The authors of the study concluded that the AI was able to take on part of the social and motivational role otherwise played by team colleagues through its linguistic interaction. For many, working with ChatGPT not only felt like using a tool, but also a bit like teamwork - including positive reinforcement.
"The study confirms our own experiences as well as those of our customers. An AI agent like AIMAX® actively ensures better work results, better performance and higher employee satisfaction.
AIMAX® even goes a decisive step beyond the ChatGPT used in the study and can not only generate texts, but also take over entire subtasks independently.
I would be happy to talk to you personally."
Like any study, this field experiment also has some limitations that should be taken into account when interpreting the results. The authors explicitly name several limitations:
- Team size: only individuals and teams of two were studied. It is unclear whether the findings can be transferred to larger teams - AI influences could have a different effect in larger groups.
- Duration & context: The task was limited to one-day innovation projects in a single company (P&G). Longer projects or other company or industry contexts could lead to different results. The generalisability of the results to all types of knowledge work is therefore limited.
- Use of technology & user experience: ChatGPT-4 (or GPT-4-based models) was used in the experiment and the participants were not AI experts at the time of the study. It is possible that more modern AI models (which have been available since 2025) or a more experienced use of prompting could achieve even stronger effects. The current results therefore represent a conservative lower limit of what is possible.
- How the AI works: The generative AI used was available in the form of a chatbot that is not specifically designed for team collaboration. Future AI systems that are better integrated into team processes could possibly act even more effectively as "team colleagues".
None of these limitations diminish the core findings, but they do make it clear that further research and experience is needed. Managers should therefore view the results in context and not transfer them to every situation without reflection. For example, more complex projects or different work cultures could reveal different dynamics between AI and teamwork.
The study impressively shows that AI support is more than just a productivity tool - in many respects, it can act as an additional team colleague. With AI, individual employees achieve the same level of performance as teams and bridge knowledge boundaries between departments. At the same time, employees find working with AI motivating and supportive, similar to interacting with human colleagues.
This is an important realisation for companies: AI can replicate the core benefits of teamwork - improved performance, shared expertise and a positive working atmosphere - and thus become a real "player" in the work process.
For decision-makers, this means that AI should no longer be viewed purely in terms of efficiency. Instead, the results suggest that work processes and team structures should be rethought in order to utilise the full potential of AI.
If AI can temporarily take on the role of a team member, organisations should consider how they design human-AI teams: e.g. in personnel planning, training (focus on effective use of AI) and in dealing with technical specialisation (AI can democratise knowledge).
Ultimately, the study emphasises that AI in knowledge work not only automates tasks, but can also fundamentally change the way in which collaboration works. This requires decision-makers to adapt corporate structures with foresight and deal openly with "AI as a team member" in order to secure future competitive advantages.
