Roadmap Unlocks Gen AI’s Value While Managing its Risks

In the rapidly evolving technology landscape, generative AI (Gen AI) stands out as a beacon of transformative potential to generate impact across innovation, growth, and productivity. This technology, capable of producing software code, text, speech, high-fidelity images, and interactive videos, is not just a tool of innovation but a potential driver of significant economic uplift. McKinsey research suggests that AI and Gen AI could add up to $4.4 trillion in economic value globally, and a productivity uplift of 3.4% annually for Mexico until 2040. 

While many corporate leaders are determined to capture this value, there is a growing recognition that Gen AI opportunities are not risk-free. In a recent survey of more than 100 leading organizations, McKinsey found that 63% of respondents characterize the implementation of Gen AI as a “high” or “very high” priority; however, 91% of these respondents don’t feel “very prepared” to do so responsibly.

That unease is understandable. The risks associated with Gen AI range from inaccurate outputs and biases embedded in the underlying training data to the potential for large-scale misinformation, cyberattacks, and malicious influence on politics and personal well-being. However, by adapting proven risk management approaches to Gen AI, it is possible to move responsibly and at a good pace to capture the value that the technology can enable.

To navigate these challenges, leaders should consider a structured approach to implementing Gen AI, focusing on understanding risks, developing governance structures, and embedding these frameworks into their operating models. Four steps can help de-risk:

Step 1: Launch a sprint to understand the risk of inbound exposures.

Start with a comprehensive assessment of inbound risks and risks that directly result from the adoption of Gen AI. In fact, it is a good practice to update risk assessments regardless of the active engagement with Gen AI as the technology can give rise to increased threats in fraud and cybersecurity (with an increased volume and sophistication of attacks from gen-AI-enabled malware, for example). 

Given the evolving nature of the technology underlying Gen AI and its applications, businesses will need to repeat the effort to identify their exposure with some regularity. For most, refreshing this exercise at least semiannually will be important until the pace of change has moderated, and the control environments and defenses have matured.

Step 2: Develop a comprehensive view of material risks and build management options.

Once potential risks are identified, organizations need to develop a nuanced understanding of them across various domains and use cases. This step involves categorizing risks according to their severity and likelihood and then crafting a suite of risk management options that include both technical and non-technical measures. 

Technical measures might involve developing new AI monitoring tools or enhancing data encryption techniques, while non-technical measures could include establishing stricter protocols for data usage and sharing. The goal is to create a robust framework that can dynamically adapt to the evolving nature of Gen AI and its applications in the business landscape.

Step 3: Establish a governance structure to support rapid decision-making.

Effective governance is crucial for managing Gen AI responsibly. Organizations should establish a structure that not only encompasses expertise and oversight but also supports agile decision-making processes. This could involve adapting existing governance frameworks to include considerations specific to Gen AI, such as special committees or working groups that focus exclusively on AI-related issues. 

These groups would be responsible for ongoing monitoring of AI deployments, assessing new risks as they emerge, and ensuring compliance with both internal policies and external regulations. The governance structure should balance the need for thorough risk assessment with the organization’s desire for rapid innovation and deployment of AI technologies.

Step 4: Embed the governance structure in an operating model that includes training.

The final step in the roadmap involves integrating the governance framework into the organization’s broader operating model. This includes aligning the Gen AI strategy with the overall business objectives (the starting point for drafting a business-led roadmap) and ensuring that all employees understand their roles in implementing Gen AI responsibly.

Training programs should be developed to educate employees about the potential risks and ethical considerations associated with Gen AI. Moreover, the operating model should facilitate collaboration across departments to foster a culture of transparency and accountability in the use of AI technologies.

Given the fast-evolving nature of Gen AI, it is imperative that organizations consider these steps and commit to continuous learning and adaptation. This means regularly revisiting and revising the risk management strategies and governance structures to accommodate new developments in the technology and its applications. Businesses should consider setting up dedicated teams to track advancements in AI and propose adjustments to the governance and operational strategies accordingly.

There is little doubt that Gen AI has the potential to redefine how people work and live. While the technology is fast developing, it comes with risks that range from concerns over the completeness of the training data to the potential of generating inaccurate or malicious outputs. Business leaders need to revise their technology playbooks and drive the integration of effective risk management from the start of their engagement with Gen AI. 

This will allow for the safe and responsible application of this promising technology, helping companies manage known risks while building the muscles to adapt to unanticipated risks as the technology's capabilities and use cases expand. With a major potential uplift in productivity at stake, working to scale Gen AI sustainably and responsibly is essential in capturing its full value.