The path to targeted and sustainable generative AI

As organizations move fast toward the use of generative AI, much has been written about common considerations to keep in mind when exploring and implementing this promising technology—considerations such as choosing the right AI solutions and provider, managing AI data including its security and privacy, using AI responsibly, and so on.

Yet, what is often missing in this dialogue are two important considerations in the journey to generative AI: knowledge sovereignty and sustainability.

Without question, both AI and sustainability are important trends across industries and geographies. CGI’s 2023 Voice of Our Clients research, for example, reveals that 57% of business and technology executives are investigating artificial intelligence or doing proofs of concept, and 55% view sustainability as highly core to creating future value for stakeholders. Both are key to creating competitive advantage.

However, many challenges come with each, including: 1) narrowing the scope of AI data, which is where knowledge sovereignty comes into play, and 2) managing AI’s impact on sustainability. Both challenges need to be considered and addressed to ensure organizations both meet the promises of generative AI and continue to advance their climate commitments.

Reigning in massive volumes of data

Data fuels generative AI, and the volume of available data is colossal and ever-growing. This raises the issue of knowledge sovereignty. How do you determine which types of data and how much data to scan, store, analyze, and use for your business purposes?

Knowledge sovereignty involves setting boundaries to guide the use of AI’s power in scanning massive volumes of data. The end goal is to compile a targeted and trusted set of data assets that are fit for purpose.

Specialized expertise and skills are required for effective knowledge sovereignty. The right knowledge management model and systems are needed as well. For example, is a decentralized or centralized knowledge management model best for your organization? What type of systems do you need to interact with large language models—and how will you choose the right ecosystem? In addition, how do you align your model and systems with regulatory requirements, such as Europe’s General Data Protection Regulation (GDPR) and other similar schemes across the globe? Finally, how do you ensure the responsible use of generative AI as you scan and manipulate massive amounts of data for new business purposes?

I invite you to read our blogs that explore more on the topics of embracing the responsible use of AI and adopting a multi-model AI ecosystem.

Ensuring generative AI is sustainable

Another key consideration when it comes to implementing generative AI is sustainability. The computing power used to scan massive amounts of data, analyze patterns, and mimic human behavior demands high quantities of both electricity and water—electricity to run the systems and water to cool them. For example, companies like Microsoft, OpenAI, and Google have reported significant jumps in their water consumption as they develop new AI solutions. To reduce the heat generated by AI systems, they have to pump water from nearby sources, like rivers, into data center cooling towers.

Organizations looking to introduce or accelerate the use of generative AI need to consider the increased energy consumption it requires and the associated costs. Ways for driving energy efficiencies are needed. This is where an expert with both AI and sustainability know-how and experience can help.

With our own clients, we perform assessments demonstrating how they can narrow their AI projects to focus on the right knowledge, i.e., ensure knowledge sovereignty, while implementing energy-efficient AI practices. These assessments can help them reduce their AI-related energy usage by as much as 50%.

Developing a targeted and sustainable generative AI roadmap

If your organization is exploring the use of generative AI or currently on an implementation journey, here are some recommendations for factoring in knowledge sovereignty and sustainability:

1. Take an incremental approach: Approach your generative AI initiatives incrementally, starting with a basic assessment of your current landscape and clear identification of requirements, opportunities, and challenges. Then, build a step-by-step roadmap, with realistic milestones and goals for what and how data is accessed and the resulting environmental implications.
2. Develop proof-of-concepts: In our AI work with clients across industries, we have learned the value of proofs-of-concept (PoC). Taking the time to complete a PoC before jumping into a full-scale implementation project, especially in a fast-evolving area like AI, significantly minimizes risks and ensures better outcomes.
3. Find the right partner: Finding a partner that can offer not only AI expertise and solutions, but also knowledge management and sustainability capabilities is key. Such a partner can provide the end-to-end guidance required for a successful implementation journey.
4. Ensure strong collaboration: The right partner isn’t of much value unless there is strong collaboration between the partner’s teams and your teams. Make sure your partner offers best practices, tools, and other resources to ensure effective teamwork.

Generative AI is changing business and the world. There is much to take into consideration when moving toward AI. The better prepared you are, the better your outcomes. We believe knowledge sovereignty and sustainability are two success factors on the journey toward generative AI. Investing in each can give you a competitive edge.

CGI is skilled in both areas and can help you maximize each. Contact me to learn more about our work.