IDC Offers a Framework for Developing a GenAI Strategy
To help organizations understand how to leverage generative artificial intelligence for business success, IDC has laid out the foundational activities associated with the technology investment, issued guidance on prioritizing use cases, and identified the key stakeholders required to build and implement successful initiatives as part of a framework it is calling the Generative AI Path to Impact.
Before any of the core technologies of generative AI are explored, IDC believes the following structures need to be in place:
• A responsible AI policy. This must include defined principles around fairness, transparency, protections, and accountability relating to the data employed to train models, as well as how the results are used. A responsible AI policy should also provide transparency on the roles and responsibilities of developers, users, and other stakeholders, while addressing legal and compliance issues.
• An AI strategy and road map. A set of defined, measurable, and prioritized generative AI use cases is required to align the organization on the key areas that will deliver the maximum business impact in the short, medium, and long term.
• An intelligence architecture. Managing the life cycle and governance of data, models, and business context for every use case is critical. The architecture should also include protocols for data privacy, security, and intellectual property protection.
• A reskilled and trained staff. New competencies will be required to build and use generative AI models, such as “prompt engineers” to write and test prompts for generative AI systems. Every organization must create a new skills map for core AI technologies and business capabilities to deploy generative AI at scale across the organization. Organizations should also build personalized training program for key roles.
Once these are in place, organizations must develop a clear understanding of the core generative AI technologies, as well as their foundation models and capabilities. At the center of any generative AI system is a generative foundation model, including the well-known large language models (LLMs). The game changer in the AI market is the ability for these models to be trained on extraordinarily large amounts of semi-structured and unstructured content and generate new content based on simple prompt requests.
The next step in defining the path to generative AI impact is prioritizing an identified set of use cases. IDC defines a use case as a business-funded initiative enabled by technology that delivers a measurable outcome. There are three broad types of generative AI use cases that need to be assessed:
• Industry. These involve more custom work and, in some cases, may require organizations to build their own generative AI models. Examples include generative drug discovery in life sciences and generative material design for manufacturing. Specialized use cases tend to be built around specific models and model providers, with custom integration architectures designed for individual clients.
• Business function. These use cases typically involve integrating a model (or multiple models) with corporate data for use by specific departments or business functions, such as marketing, sales, and procurement. Many organizations are already testing these types of use cases but are concerned about intellectual property leakage and data governance.
• Productivity. These use cases are aligned with work tasks, such as summarizing reports, creating job descriptions, or generating Java code. Generative AI functionality for productivity improvement is being infused into existing applications, such as Microsoft 360 Copilot or Duet AI for Google. For many of these use cases, business value can be delivered through the content and data on which the underlying foundation models have been trained.
Ultimately, generative AI will be widely adopted only if the data, models, and applications that use it are trusted by end users and customers. To achieve this, organizations need to establish a well-orchestrated trust and oversight program to ensure that generative AI technologies can be deployed in a sustainable manner. Organizations and AI vendors must understand the benefits and limitations associated with generative AI use and be prepared to remediate issues while complying with regional data privacy regulations.
Finally, IDC recommends adopting a three-part framework to help organizations transform their business models using generative AI. The first part focuses on near-term, incremental innovation. The second is disruptive innovation in the medium term. The third focuses on long-term business model transformation. The framework drives alignment across all business domains and helps prioritize key initiatives.
“As the industry moves forward with this fundamental transition to AI embedded into every business and technology function in the enterprise, IDC believes that every CEO will need to have an AI strategy, and generative AI is the trigger,” says Phil Carter, group vice president of thought leadership research at IDC. “It is best to get started quickly. We are hopeful that this framework will help every organization develop their own path to impact.”