As the hype about AI reaches near-frenzy levels, many organizations are wondering about what they should do in order not to be left behind by competitors who are doing better at utilizing newly developed technologies, and those introduced almost every day.
AI is not new, and the quest for machine intelligence started in the 50’s of the last century with the question as to whether machines could “think” in the seminal paper written by the father of modern computing, Alan Turing. It must be noted that there have been a few AI “winters” or “false dawns”, though. In the 70s, there was the prediction that machine intelligence was imminent but didn’t materialize. In the 80s, the grand promises of applied artificial intelligence applications like Expert Systems, which used logic programming applied to hard-coded knowledge, again didn’t materialize. In this short piece, I share my thoughts on what modern organizations can do to leverage new technologies and reduce the threat of new disruptors that could leave them suffering a fate similar to that of Nokia’s and Kodak’s.
The current remarkable success of generative AI (or GenAI for short) systems, such as ChatGPT, Gemini and Claude, to name but a few, has spawned a variety of use cases in businesses of all kinds and in manufacturing organizations as well. However, organizations wishing to benefit from GenAI should realize that technology always comes with limitations. For instance, GenAI models have built-in biases. Furthermore, they need careful and well-structured prompting to deliver useful results. They may also leak confidential data, and could sometimes generate false statements.
In any case, organizations should carefully consider their responses to this exciting new technology. GenAI should be viewed as an enabling technology that must be deployed and used wisely. In my view, the best response would be a strategy that fits both the organization and the modalities of these GenAI tools. Accordingly, I believe that a useful GenAI integration strategy would keep the following in mind:
Organizations should start with a clear positioning of the role of data and knowledge in their business strategies and how they support those business strategies. We often speak of algorithms and forget data, which is a crucial parameter of any successful AI application in business. They should ask themselves what role information and knowledge systems are expected to play in operationalizing or supporting their strategic thrusts, or, in other words, what possible roles GenAI tools, or any evolving AI technology, could perform for them. The strategy could be presented as a set of strategic scenarios where the play of IT, in whatever shape or form, could support critical business value chains.
Since most AI tools are based on either public or private knowledge, there should be an organizational knowledge audit that generates the fundamental and critical knowledge structures in the organization. Large Language Models (LLMs) that lie at the heart of text-based generative AI tools need to be fine-tuned to the specific business domains that they are aimed to serve. This requires that organizations have models of their own knowledge assets. Such knowledge models could be “architectonic knowledge models” that clearly differentiate between core knowledge, that is relatively more stable and change more slowly, and other types of knowledge that are less stable and are subject to frequent and rapid changes. Organizations require explicit, easily-shared models of the knowledge they need in order to excel in delivering the products and services that are crucial to their operations. Such knowledge should be a good base for any future AI exploitation endeavours.
Moreover, organizations need clear strategies and mechanisms for leveraging knowledge outside of their boundaries. This could be in the form of data that pinpoint customers’ pain points, desired product features and attributes, as well as ideas for new products and services. A robust approach to exploiting generative AI would have processes to distil useful information from publicly available data such as product reviews, customer complaints, warranty claims, and blogs about the products or services.
Organizations also need to immediately start building an internal focus of knowledge about adapting and using generative AI tools, perhaps by designating and training “AI Champions”, either as individuals or in groups, who can become internal nodes of expertise. They should create for them the climate they need for further learning. That could be done through providing them with their own “sandboxes” for playing and experimenting with GenAI. They should also encourage all staff to brainstorm ways in which GenAI can support their operations. This would, in my view, require a certain approach to the spatial design of workspaces in a way that encourages unplanned encounters and informal conversations, as well as the exchange of knowledge and ideas.
It is also highly recommended to take a more “extrovert” approach in seeking to leverage the use of GenAI. Joining local and industry-specific interest groups, and those in academia and research centres, are some of the ways to make this happen. The fact that many universities and educational establishments seek to place learners and researchers in industrial internships of many sorts should also not be neglected, as they could serve as a bridge that connects organizations to diverse types of useful expertise and perspectives.
Finally, GenAI adoption and integration strategies need to pay close attention to designing the interactions and experiences between AI systems and humans. Any technology that is deployed in an organization becomes part of a human activity system that will inevitably influence and be influenced by the human and social contexts, so this impact needs to be carefully scrutinized, and the deployment wisely planned.
This article was written by Galal H. Galal-Edeen, Professor of Information Systems Engineering, Department of Management, AUC School of Business
