GenAI isn’t the revolution—How we use it is
While everyone talks about GenAI, we focus on humans. The real challenge of AI-driven innovation is not just about implementing the latest tools, but about understanding how people with different backgrounds, roles, and objectives reinterpret and personalise them in ways that redefine organisations and uncover innovative use cases. GenAI is not a static technology—it is constantly appropriated, adapted, and redefined by users, often in ways its creators never imagined.
The pallet effect: When a simple tool becomes something more
To understand this, consider the wooden pallet. Originally designed for logistics and transportation, pallets started out with a clear industrial function. Yet over time, people reinterpreted them—transforming them into furniture, beds, coffee tables, and even sustainable design pieces.
No manufacturer planned for pallets to become a symbol of upcycling and sustainability—this meaning was created by users. Similarly, GenAI was designed for specific tasks like text and image generation, but its true potential is shaped by the diverse ways people appropriate it. Some use it to write legal contracts, others to compose music, while companies leverage it for software development, customer support, and even scientific research.
At PwC, our Copilot Transformation Project has demonstrated this exact process of reinterpretation. What began as a tool for enhancing quality and productivity has evolved into a catalyst for cross-functional learning, process redesign, and innovation management transformation. The project was not just about ensuring a smooth and effective adoption of the technology, but also about experimenting with it to discover innovative use cases. Our goal was to determine how and where Copilot could be integrated into workflows, team dynamics, and business processes, and whether its potential extended to other contexts.
Just as the pallet’s meaning depends on its social and cultural context, so does GenAI’s impact.
The key question for organisations is: Are you fostering a culture where GenAI can be reinterpreted, adapted, and used meaningfully?
Innovation isn’t about using AI—It’s about reimagining It
When a new technology enters an organisation, it doesn’t just get implemented—it gets reshaped by existing structures, organisational habits, and individual behaviour. This idea is central to Adaptive Structuration Theory (AST), which explains how new technologies interact with and change established work environments. As DeSanctis & Poole (1994) explain, AST suggests that technology adoption is not a linear process but rather a dynamic, bidirectional interaction: technology influences the organisation, and the organisation influences how the technology evolves in practice.
The Copilot Transformation Project exemplifies this theory in action. By integrating Copilot into PwC’s business units, we observed that different teams appropriated the technology in unique ways—from automating repetitive tasks to developing entirely new ways of working. This reflects what Greco & Corvello (2024) describe in their research: “Managers who received more comprehensive training on how to customise and interact with AI tools were more likely to integrate GenAI into their workflows in ways that enhanced productivity and innovation.” At PwC, we recognised this early and ensured that our Copilot deployment was accompanied by hands-on guidance, upskilling sessions, and best practice sharing across different PwC territories.
Question for leaders: Are you enabling your teams to experiment and make AI their own, or are they simply following pre-set rules?
The AI trust equation: When to believe the machine
A common misconception about AI is that its outputs must always be 100% accurate to be useful. In reality, organisations should be asking: What level of trust is required for different AI-driven tasks?
As Piller & Srour (2024) explain, “The issue is not to find a way to make all AI-generated results accurate so that they are always trustworthy but rather to be able to assess the level of trust that specific use cases require and adjust the accuracy of GenAI results accordingly.”
At PwC, we understood this early on with our Copilot deployment. Some processes, such as drafting emails or summarising meetings, require speed and creativity rather than absolute precision. Meanwhile, for financial and regulatory tasks, accuracy and compliance are paramount, necessitating careful oversight and human validation.
Question for leaders: Are you distinguishing between AI applications that require flexibility and those that demand absolute trust?
PwC’s Copilot Transformation Project: From adoption to transformation
At PwC Luxembourg, the rollout of Microsoft 365 Copilot to 1,500 employees was more than a technology deployment—it was a strategic shift in how AI and humans collaborate.
- Breaking down silos: The project encouraged collaboration across different PwC territories, leading to best practice sharing and co-creation;
- Fostering human skills: We ensured that technical teams worked with designers and humanities experts to bridge the gap between AI and business; and
- Measuring real impact: A 93% adoption rate showed how the tool became embedded in daily work, but more importantly, we observed a shift in how people worked together.
To achieve this, PwC adopted a multi-phase approach:
- Training and upskilling sessions to familiarise employees with the tool;
- Best practice sharing across PwC territories to exchange insights and learn from early adopters; and
- A mini-hackathon where IT and business teams will collaborate to experiment with Copilot agents, solving specific pain points in business workflows and projects.
The lesson? GenAI adoption is not a plug-and-play solution. It requires organisations to rethink how humans and AI agents collaborate.
Question for leaders: Are you treating AI adoption as a one-time rollout, or as an ongoing evolution of work?
When AI starts talking to itself—What happens next?
One of the most profound shifts AI is bringing to work is the redefinition of roles. Employees are not just using AI anymore—they are supervising and collaborating with AI agents in ways that fundamentally change workflows. But what happens when AI starts working without human supervision?
As Cimino et al. (2024) put it, “One of the most profound effects of GenAI is its impact on the structure of work. Traditional roles are being redefined, with employees increasingly required to manage, supervise, and collaborate with AI systems.” But what happens when AI starts collaborating with itself?
At PwC, we see Copilot as the first step toward this shift. But in a world where AI agents begin to interact autonomously, optimising their processes and decision-making logic, how do we ensure that humans remain at the centre of innovation?
We’ve talked about how humans reinterpret AI. But what happens when AI starts reinterpreting itself?
A question worth exploring—before the agents answer it for us.
For more insights on design-driven innovation, see my previous article: Design-Driven Innovation: How to Reshape the Way You Do Business Through Real Use Cases.
