Care and dare: How pharmaceutical leaders can empower AI experimentation without losing control

AI is not new to the pharmaceutical industry. For years, it has played a role in areas like drug discovery, preclinical research, and data analysis. What is new is the scale and reach of AI today. It is no longer confined to specialist teams or isolated use cases. It is touching almost every function, from R&D and clinical trials through to manufacturing, supply chain, and commercial operations.

Yet despite this growing presence, many organisations are struggling to realise the transformative value AI promises.

The issue isn’t awareness, investment, or even activity. It’s that AI is often being absorbed into existing ways of working rather than being allowed to change them. And that is fundamentally a leadership and cultural challenge.

One of the most important shifts we explore in our whitepaper, Beyond the Algorithm: How Pharmaceutical Leaders Can Navigate Cultural Transformation in the Age of AI, is how leaders enable experimentation and learning without undermining the rigour the industry depends on.

This is where the idea of “care and dare” leadership becomes essential.

Why AI experimentation is often constrained

The pharmaceutical sector is built on precision, rigour, and control. For good reason. Patient safety, regulatory compliance, and scientific integrity demand it. But AI introduces a very different dynamic.

AI thrives on iteration, learning, and trial-and-error. More often than not, value doesn't emerge from perfect plans, but from a rapid process of testing, failing, adjusting, and trying again. This creates a natural tension with long-established ways of working in pharma, which traditionally operates more slowly and cautiously.

If this friction isn’t actively addressed, AI initiatives tend to stall. Teams experiment warily, learning remains localised, and promising ideas struggle to scale beyond pilots. The organisation stays busy, but breakthroughs remain elusive.

What’s needed is not a loosening of standards, but a more deliberate leadership stance on where precision is essential, and where learning must be accelerated.

Introducing "care and dare" leadership

In our work with pharmaceutical leaders, we increasingly see the need for a clear distinction between two modes of operating:

Care: In core business areas (manufacturing, quality, regulatory processes, supply chain) operational excellence remains critical. These are environments where mistakes carry real cost and consequence, and where consistency and control must remain high.

Dare: In AI-enabled innovation, leaders must create space for experimentation. This includes testing new ways of working, learning what doesn’t work, and iterating quickly. Here, progress depends on exploration rather than perfection.

The mistake many organisations make is trying to apply the same mindset to both. Care without dare leads to paralysis. Dare without care creates unacceptable risk. The leadership challenge is holding both at once, intentionally.

Giving people the tools and permission to play

One of the most powerful examples we highlight in the whitepaper comes from Roche.

As Bryn Roberts, SVP & Global Head of Data, Analytics & Research, puts it: "Beyond our deeper scientific and technical use-cases, we encourage 'everyday AI' through playful exploration by everybody. Once people appreciate what's possible, a little training and a safe environment enables them to 'go and play' in their daily work."

Instead of tightly prescribing use cases, leaders provide secure and approved tools, clear guidance on what not to do, explicit permission to experiment. And then they step back.

This approach recognises that you can't learn what works by theorising alone. Learning emerges through use, through friction, and yes, through failure.

As Bryn also notes: "It's only by failing that we're actually going to see what works and what doesn't."

Guardrails matter more than ever

However, empowerment doesn’t mean the absence of structure. As AI becomes more embedded in daily work, leaders must be explicit about where experimentation is encouraged and where constraints apply. This includes clear expectations around data use, privacy, and compliance, and strong governance that supports learning rather than suppressing it.

When these boundaries are clearly communicated people stop second-guessing what’s allowed and psychological safety builds. Teams are able to experiment responsibly, rather than cautiously or covertly. In other words, guardrails enable learning,

From supervision to enablement

This shift also requires a change in leadership behaviour. AI-enabled teams don't need leaders who monitor every task. They need leaders who coach rather than control, frame experimentation as learning, not performance, and who share their own uncertainties and lessons learned.

When leaders openly engage with AI themselves and share what they’re learning, they signal that experimentation is not only safe, but expected. This builds the psychological safety that allows teams to move faster and learn together.

Why the shift matters now

AI is already reshaping how work gets done. People who learn to work effectively with AI are becoming dramatically more productive. Teams that experiment responsibly are moving faster than those waiting for certainty.

As our whitepaper makes clear, AI will not replace people. But people who use AI will very quickly replace people who don't.

The organisations that succeed won't be the ones with the strictest controls. They'll be the ones whose leaders can hold the tension between care and dare-protecting what must be protected, while deliberately creating space for learning, experimentation, and growth.

That balance isn’t easy, but it is learnable. And it's fast becoming a defining leadership capability in the age of AI.

Want to explore this further?

Our whitepaper, Beyond the Algorithm: How Pharmaceutical Leaders Can Navigate Cultural Transformation in the Age of AI, dives deeper into the leadership shifts, cultural conditions, and practical behaviours required to make AI work at scale. Download it to explore how to empower your people without losing control.