Effective guardrails transform gen AI fascination into solid foundations

What is gen AI, and how is it driving change?

ChatGPT reached 100 million users within two months of launch. The story of generative AI (gen AI) is a story of technology that has very quickly gained massive popularity, and a story of technology that is very quickly evolving. Some eagerly embrace gen AI while others voice privacy, cybersecurity, ethical and other legitimate concerns. Look no further than 2023’s labour dispute between Hollywood creatives and studios, which was punctuated by a host of issues — including the use of gen AI to create digital doubles of actors, alter facial expressions or voices and generate entirely new dialogue or scenes. The 148-day strike between the Writer’s Guild of America (WGA) and Hollywood studios only ended when the union successfully negotiated desired limits on the use of AI.

What lessons can we learn from this very impactful, very human reaction to the awesome power of generative AI?

 At EY, we’ve been enabling clients to embrace AI technologies at scale for years. Gen AI is already infused across our key practice areas, from risk modelling to data modernization to cybersecurity. We have learnt that a balanced approach to AI governance promotes innovation while mitigating risks. It gives all stakeholders a voice in deciding how technology is used. Ultimately, this creates a healthy information ecosystem that’s likely to optimize better outcomes for everyone while incentivizing new ideas that drive economic growth.

Tapping into gen AI’s potential requires enterprises to rethink business strategies and operations by putting humans and machines together, at the very centre of a responsible and values-driven approach. At every stage of this evolution, it’s remained consistently clear that making the most of this technology depends on an organization’s ability to do so while implementing strategic guardrails along the way.

While other forms of gen AI include technologies that can code or create images, among other tasks, Azure OpenAI’s ability to support businesses is generating the lion’s share of attention. We see organisations generating new product designs created by AI. We see better executive decision-making supported by improved real-time and good-quality data. We see improved employee and customer experiences thanks to cleverer digital assistants. All this and more is now possible within the workplace using the OpenAI service inside Microsoft’s Azure – leveraging the same awesome power behind the wildly popular ChatGPT, but with enterprise grade security & compliance.

Still, the question remains: how can we move from the current stage of fascination to one of responsible and value-driven generative AI implementation while prioritizing human wellbeing?

Why are guardrails so important to effective gen AI adoption?

In the absence of proper controls, AI adoption may expose the organization to potential failures and mishaps. EY teams early experience with AI with the clients emphasizes the need for sound governance and risk management to safeguard against risks tied back to design, data, algorithms, performance, third parties, regulatory requirements, reputational issues and business processes – just to name a few.

While eliminating risk entirely may seem impossible, guardrails can certainly have a considerable impact. That said, the most effective gen AI guardrails emerge through a carefully constructed trifecta that bridges:

• Processes, such as model risk tiering, two lines of defence, operating model, ethics/privacy assessment)
• Technology, including data lake, cloud computing infrastructure, vendor tool management
• People, including leadership oversight, talent management and training, policies and role guidelines

The EY organization is taking a global leadership role in helping the clients with the responsible rollout of gen AI as part of our larger commitment to help shape AI that builds a better working world.

"We support institutions in their AI adoption journey through the build-out of an AI governance framework and operationalization organization wide. Our accelerators help institutions build their AI policies and risk management framework and put it in practice to mitigate business, reputational and regulatory risks. We have experience in this domain with top banks in North America and other non-financial institutions." – Yara Elias, Senior Manager, Consulting, Ernst & Young LLP (EY Canada).

How can organizations prepare to adopt gen AI at scale?

Organizations around the world are wrestling with the seemingly urgent need to implement gen AI — even though the business may not feel ready to dive in. The first step in adopting and implementing this new tool is determining where across the business gen AI can enable value creation. Whether you surface opportunities in finance, legal or tax through to risk, customer service or supply chain: grounding your gen AI approach in the potential for value creation is essential. This supports a confident, transformative approach that’s vital to unlocking truly exponential value.

A connected approach can help you cement that foundation quickly and effectively. Here at EY, we embrace a three-staged process based on our:

• EY.ai Maturity Model. Helps you ideate growth opportunities for enterprise-scale AI adoption, evaluates your market position and gives you an understanding of your ethical compliance.
• EY.ai value Framework. Supporting the maturity model, this helps you prioritize your chosen gen AI initiatives so you fund the ones that drive the most value while controlling sprawl and helping you be inclusive in your use case intake process.
• EY.ai Confidence Index. Runs an empirical assessment of the inputs and outputs of an underlying AI model, and then integrates ethical, societal and public policy considerations. This helps shape your own responsible AI guidelines and frameworks while giving you confidence in the AI you create.

This integrated process is working across the globe.

The EY-Microsoft collaboration has been accelerating customers towards real-world implementations of gen AI.  We recently built a gen AI tool to automatically create product detail pages for a global consumer packaged goods organization, including customer voice retraining and retailer-specific optimization. Results included optimized paid media spend and boosted organic rank and conversion.

With a major telecommunication company, the team designed and deployed a customer care transformation program. This undertaking included hyperautomating processes, speech and advanced analytics, document intelligence and intelligent virtual agents.

In another case, we integrated Azure OpenAI into the core platforms of a global financial institution, which will improve advisor and associate access to internal text, voice and image content — positively impacting the employee experience and improving the overall customer experience. 

Even so, none of these gen AI implementations can reach their full potential without the right guardrails to marshal progress, strengthen privacy and bolster cybersecurity.

“EY and Microsoft are working closely in Canada to develop an approach to support Canadian businesses to onboard this new technology and integrate it into their overall modernization programs. Our approach builds in the necessary governance and compliance framework and addresses core data requirements to support the safety and sustainability of gen AI. By harnessing the power of generative AI in a purpose-driven manner, together we can unlock new dimensions of creativity, productivity and innovation that benefit individuals, organizations and society as a whole.”

-Melissa Tamblyn, Microsoft Practice Leader, Ernst & Young LLP (EY Canada)

"In the era of generative AI, where OpenAI and Microsoft are paving new paths, being a mere follower is akin to standing still in a digital marathon. To truly capitalize on the transformative potential of AI, one must not only initiate action but also construct it upon a foundation of ethical guardrails. It's about striking a balance between innovation and responsibility, ensuring that our strides in AI yield tangible ROI and add value not just to our customers, but to our employees as well. Our partnership with EY allows to meet Customers where they are and get started. By embedding the right principles and responsible AI practices from the get-go, we turn novelty into legacy. "

- Gerardo Amaya, Senior Industry & Tech Leader for Global Advisories - Americas Region

What’s the bottom line?

As the world continues to embrace the potential of gen AI, ensuring that its implementation is carried out ethically, responsibly and with the human element at its core is critical to achieve sustainable value and long-term success.