30 Institutionalizing Gen-AI Versatility Gen-AI tools are still in their nascency, meaning their capabilities and possible use-cases will expand significantly over time. For example, compound AI systems, multi-agent AI systems, and Artificial General Intelligence,33 are all adjacent to Gen-AI tools but are still emerging fields at the time of this publication. These tools and others could quickly expand use-cases for Gen-AI tools in unpredictable ways. Further, as workforces adapt to using these tools, new applications for existing capabilities will emerge. Organizations, therefore, need to account for both the expansion in applicability and technological capability. To do so, we recommend developing internal operating norms and structures that foster Gen-AI versatility over the long run. Below we outline potential approaches to institutionalize versatility within an organization as it integrates Gen-AI tools: Establish Streamlined Reporting Systems Developing channels to identify, report, and circulate new uses and adaptations of Gen-AI tools within an organization can upskill the entire labor force. Organizations will want to consider establishing a formal reporting system that uploads use-cases, successful prompt examples, and general tool knowledge to existing company data repositories, such as enterprise intranets or knowledge sharing applications. Such a reporting system can be implemented through existing enterprise tools, or through an extension of the Gen-AI tool itself, though this will depend on the vendor’s capabilities. Ease of access and ease of use for these systems can help streamline the reporting process and prevent administrative burden from stifling usage. Further, incentivizing adoption can encourage employees to consistently use the system. Specific incentives will vary between organizations but should be designed to encourage both tool adoption and reporting through the system. While building use-case repositories and incentivizing the workforce to leverage them is a logical starting point, Glenn Parham, former leader of CDAO’s Generative AI-focused Task Force Lima, cites the need for more granular data and analysis to enhance reporting systems. Appoint Ambassadors Tasking leaders in an organization with incubating, managing and improving Gen-AI tool usage across the workforce can help provide structure, accountability, and curation for reporting systems. Enterprises with sufficient resources can hire in-house teams versed in a wide range of tools and their implementation. A lower-cost approach involves tasking existing teams broadly aligned with technology development and adoption, such as Offices of the Chief Innovation Officer, Chief Technology Officer, Chief Data Scientist, or Head of R&D. One example of a military successfully implementing a structured operational model conducive to AI incubation is the Israeli Defense Forces (IDF). Specifically, the IDF has a subset of its organization designated for the incubation and development of emerging technologies, with AI as a major area of focus. The IDF calls these teams AI Factories, which are comprised of experts in subject matter-relevant academic fields, defense industry markets (including startups and mature firms), as well as military officers. Each team assigns goals toward an operational or technology challenge, and then is granted a development period to incubate and experiment with a technology to identify applications for the military. The process, in some aspects, reverses the requirements-to-acquisition process by prioritizing experimentation with a technology before a capability gap has been identified. Such an approach helps discover breakthrough technology use-cases that otherwise may have gone undiscovered. I believe the only way to build a reliable & accurate AI inventory is by analyzing user logs (i.e. chatbot logs), and clustering unique use cases. With this approach, you get an empirical understanding of how the workforce is actually using these tools, not what leadership thinks you want to hear.34 “ Generative AI Adoption in the US Military
Generative AI Adoption in the US Military Page 29 Page 31