AI as a Full-Fledged Colleague

“Our results demonstrate that AI can replicate and even enhance many of the benefits of human collaboration.”

This conclusion follows from a fascinating, large-scale study by Harvard, Wharton, and Procter & Gamble (P&G), conducted in 2024 but published in late March 2025. While many organizations primarily view AI as a tool to automate human tasks and thus save on labor costs, the results of this study show that this approach is not only limiting but potentially even counterproductive. But what exactly does it mean when AI is not just a tool, but actively participates in teamwork? Can legal teams also benefit from AI as a full-fledged colleague? And what are the concrete consequences of this new collaboration for organizations?

The Study

In the summer of 2024, 776 experienced professionals from P&G participated in a unique experiment that reflected their actual work practice: developing new product ideas. Normally, two specialists brainstorm together – a commercial expert and a technical (R&D) specialist. For the study, this way of working was adapted, and AI was added as a colleague.

The researchers then examined four different scenarios:

• Individuals without AI.
• Individuals with AI (GPT-4).
• Teams (two people) without AI.
• Teams (two people) with AI.

Professionals working with AI used GPT-4. Prior to the experiment, they received a one-hour training session in which they were instructed on how to use the AI effectively. During this training, they were also provided with recommended sample prompts and tips for effective interaction with GPT-4.

During the study, participants primarily used AI for:

• Generating initial ideas and potential solutions.
• Testing and refining their own proposals through feedback from GPT-4.
• Combining ideas from different specializations (technical and commercial) into integrated solutions.
• Improving the speed and completeness of their work.

The AI was not simply used as a “copy-paste” tool; participants engaged in an active dialogue with the system. All participants were given real-world assignments within product categories such as personal care, oral care, and baby products.

Their proposals were evaluated by experts with relevant commercial and technical experience.

They assessed each proposal based on three key criteria:

• Quality: The overall quality of the solutions was rated on a scale of 1 to 10, based on how well the solutions aligned with business objectives and usability.
• Novelty: The originality and innovativeness of the proposed solutions were assessed.
• Feasibility: Experts assessed how practically viable and realistic the solutions were within the company.

The scores were then standardized to allow for comparisons between groups (individuals with and without AI, and teams with and without AI). Additionally, the time participants spent developing their ideas and the level of detail (word count) of their final proposals were measured.

AI Enhances Individual Performance

Not entirely unexpectedly, individuals without AI performed the worst. Teams without AI performed better, but the real surprise was that individuals with AI actually outperformed traditional two-person teams without AI. Furthermore, individuals with AI performed almost as well as teams that also used AI. This study convincingly demonstrates that AI does not simply replace people, but rather can elevate individual professionals to new heights.

In addition, the use of AI yielded significant efficiency gains. Individuals with AI completed their assignments an average of 16.4% faster and produced more detailed and well-thought-out solutions. Teams with AI finished 12.7% faster than teams without AI.

Breakthrough Ideas Thanks to AI and Collaboration

Perhaps the most remarkable outcome: teams with AI produced ideas that ranked in the absolute top (the best 10%) three times more often. This underscores a crucial added value of AI as part of human teams. It enables professionals to think outside their usual frameworks and develop exceptional solutions.

The Disappearance of Professional Silos

Without AI, professionals clearly stayed within their specialization: commercial experts primarily devised market-oriented solutions, while technical experts produced more technological ideas. AI convincingly broke down these walls. Individual professionals using AI came up with balanced, integrated solutions in which technical and commercial elements merged harmoniously. This shows how AI can function as a powerful bridge between different areas of expertise.

As the researchers aptly state:

“AI functions not only as a source of information but also as an effective mechanism for crossing boundaries between domains and enabling professionals to approach problems more holistically.”

Image from the LinkedIn page of Conor Grennan

Why This Is Important for Legal Professionals

For legal professionals in particular, these results offer important insights. Legal work requires precision, in-depth specialist knowledge, and intensive collaboration between different areas of expertise. The findings from Harvard and Wharton suggest that legal teams have much to gain by viewing AI not just as an automation tool, but as an active partner. This can be especially beneficial for young or less experienced lawyers, who, thanks to AI, can perform complex legal analyses more quickly and reach the same level as their more experienced colleagues sooner.

Daring to Think Beyond Automation

While the immediate temptation to use AI for quick cost savings is great, researchers emphasize that organizations are missing opportunities by doing so.

Ethan Mollick (Wharton) rightly states:

“Organizations currently see AI primarily as a productivity tool, like a better calculator or spreadsheet. But employees are actually using AI for critical thinking and complex problem-solving, not just for routine tasks.”

Mollick encourages organizations not only to automate existing processes but to think fundamentally differently about work structures and collaboration.

Other Practical Examples Underscore This

These results do not stand alone. A recent study within the Pennsylvania state government also confirms how AI offers concrete benefits: employees reported significant time savings and improved quality in various tasks, such as document management and policy development. The participants described AI as an extremely reliable and resourceful colleague, not simply as a tool.

Conclusion

Generative AI is forcing organizations to revise traditional ideas about teamwork and expertise. The Harvard study shows that AI is not just a means of automation but can be an active and valuable colleague. The researchers deliberately use the term “cybernetic teammate,” referring to the term “cybernetics,” coined by Norbert Wiener in the 1940s. Cybernetics is about systems that adapt through feedback loops—systems that actively respond to changes and dynamically improve themselves through interaction with their environment.

In the context of the study, this means that AI does more than simply automate tasks. Generative AI systems like GPT-4 actively adapt during interaction with humans: they not only generate information but also respond to feedback and actively help to refine ideas, integrate knowledge, and approach problems more holistically. It is therefore about AI as an active participant in collaboration, offering similar benefits to human colleagues in terms of feedback, knowledge exchange, and joint problem-solving.

Legal organizations willing to embrace AI in this way can gain a significant advantage. Not only by working more efficiently, but especially by developing new, creative, and groundbreaking legal solutions.