AI Adoption and Organisational Transformation: Balancing Innovation, Ethics, and Risk

Week of Integrity interview series

Making Integrity Tangible in a Digital Age: Concrete Practices, Insights, and Reflections for Fostering Integrity in Professional Settings

Marianna: What are the most common barriers organisations face when scaling AI across the enterprise? Is the biggest challenge in AI adoption technological, or is it preparing people for change?

Matthew Doyle:

There is so much talk on AI that many organisations, and individuals for that matter, are at a stage in their AI journey where they are driven by a mindset that, “if everyone else is using AI, we better use it as well!”. This is an example of lack of strategic intent, which I think could be described as a barrier amongst many where there is no clear path to production. Using AI, and experimenting with it, is great but there should be a defined problem to solve and strategy behind its use. There are many barriers organisations face but another that often comes to mind is cultural resistance (or workflow disruption), where employees are resistant to change where a tool may break their everyday routine or habit without clear benefit. I refer to this in my compliance training where I give the example of my music technological transition from cassettes to CDs, and eventually MiniDiscs. With every shift, there was initial resistance before embracing the change and seeing its value (though like the MiniDisc, some AI tools won't survive long-term, which is all part of the learning process). Ultimately, driving adoption is not just about rolling out new technology; it is about showing people how it actually makes their lives easier. I think we need both a clear strategy and a human-centered approach to change. AI adoption is no longer a technical challenge - the technology problem has largely been solved, so now it is an organisational and cultural problem that needs solving. Preparing people for change is another matter. In my role as an ethics and compliance professional, I focus a lot on this aspect. Humans, in general, are naturally resistant to change. It is a psychological response and manifests itself as ‘loss aversion’ - where the fear of losing what we have outweighs its potential benefits. This is ok, but we must recognise it and take steps to encourage risk taking - not what you would typically want to hear an Ethics and Compliance Officer say!

Marianna: How can organisations strike the right balance between rapid AI innovation and responsible risk management? What ethical considerations should businesses prioritise when deploying AI at scale without limiting innovation? 

Matthew Doyle:

My organisation is all ‘onboard’ with rapid AI innovation, and I intentionally use the phrase ‘onboard’. We operate what is the ‘jet ski’ approach - an idea that teams and ideas move fast, and rapidly change direction when needed, eventually steering a path towards becoming a larger and more stable boat or ship on its intended course. Many organisations use the Formula 1, or race car analogy, which is similar. This can be applied to AI innovation, however when it comes to jet skis, as they are so agile and move direction so quickly, I think of myself as riding on the jet ski itself, while providing navigation and steering advice to the driver. It is also considerably more fun this way!

When scaling AI, apart from consideration of the usual technological constraints, data access controls, etc., true ethics should focus on real human impact. Through using AI, and promoting AI knowledge, we upskill our employees. In this manner, we ensure they are valuable critical thinkers alongside the technology. It is possible thereafter for us to fast track the low risk ideas while also protecting high stakes decisions. This gives teams the green light to innovate rapidly, knowing they have clear guardrails to handle the technology at full speed.

Marianna: To empower employees, strengthen AI literacy, and build a responsible AI culture, how can businesses help employees build confidence in using AI? 

Matthew Doyle:

If it's available, people will be using it. The technology is widely available, and the majority of people, whether fully aware of the fact or not, are already using forms of AI in their private lives. This inquisitiveness—an outstanding feature of human nature—will naturally spill over into their professional roles. Employees will use AI, it is a fact we cannot deny, so businesses must take the lead in providing the right tools for them to do so with necessary safeguards in place. The alternative is that employees will naturally seek out other public source models and software putting at risk data governance, commercially sensitive information, etc. By providing reliable, and vetted, tools or software for our employees, we provide this space. It reminds me of a playground analogy I can relate to with my own children. Young children naturally want to play, so as communities and caregivers we provide dedicated spaces for them to do so - the safeguards in place being fences separating the space from the road, or to prevent children from wandering away- but within the space, children are largely left to their own devices; helicopter parenting aside! Expanding on this analogy, how do we build confidence? We encourage children to get their hands dirty, talk to other children, invent games; the same holds true for building confidence in using AI. In my organisation, we do this by encouraging experimentation, exploring new ideas, and allowing failures, within this space. We are fortunate to have a secure, closed-loop enterprise AI workspace that connects to existing tools, builds what you need, and keeps everything inside. I recognise though that this is not always within the capabilities of every organisation, but close alternatives or similar licensed tools should be considered. Otherwise, the alternative is ‘shadow IT’: employees using public, insecure tools with no governance, no audit trail, and no data control.  

Marianna: How can organisations encourage experimentation with AI responsibly without compromising governance, compliance, and trust?

Matthew Doyle:

Now that we have established that employees WILL use AI, I would like the segue into this question and I think there are two parts; 1. Encouraging experimentation, and 2. Without compromising governance, etc. Encouraging experimentation was referred to above, it is the mindset which should be driven by senior leadership of allowing employees to try new ideas and be accepting of the fact that many ideas, agents, or apps they build may not work. Along the way though, they learn and familiarise themselves with the technology. I build agents and apps all the time, and I can proudly say that three out of five do not make it past day one, and four out of five do not survive to see the weekend. Those ideas that do persist need to be hosted in a platform or manner which allows them to be shared, therefore fostering a culture of experimentation, which drives learning, and ultimately meaningful results. Despite waxing lyrical about experimenting with AI, I actually work in the legal and compliance field and must take governance seriously. In order to ensure we do so without comprising governance, compliance and trust, we have to consider what is referred to as ‘compliance by design’. This is the concept that systems and processes are created with controls from the start, rather than as an afterthought. The advantage of this approach is that it is proactive, scalable, and low friction - without allowing easy bypasses. How to achieve this though? It is driven by system architects, leadership, product owners, but also by legal and compliance professionals like me. We do this via active participation in AI agent design and solutions. Personally, I consider it my responsibility to be as familiar as I can with AI solutions, through building my own apps, using the systems we have access to, and being ‘inquisitive’ to guide my compliance decisions and ensure I am a valuable partner to my colleagues.

About author:

Based in Amsterdam, Matthew Doyle is a Group Senior Ethics & Compliance Manager focused on strategic risk, portfolio governance, and policy implementation - overseeing and advising Prosus' portfolio companies on their compliance programmes. He brings a background in regulatory compliance across global investment and development banking sectors with a dedicated focus on AI governance and implementation.