© Adobe Stock
AI systems promise great advantages. Often they deliver to promise. But some AI projects fail, ending up not being successfully integrated into organizations. What are reasons for failure? What are success criteria for the successful deployment and operations of AI systems in companies?
Prof. Yvonne Hofstetter summarizes what she has experienced working with companies. She is CEO and co-founder of the start-up 21strategies. The AI-based technology of 21strategies changes the way companies make decisions under uncertainty by using algorithmic planning and execution strategies.
Success factor One for successful integration of tactical AI systems: The business case
Companies expect to gain strategic advantages from the use of tactical AI systems. They want to prepare their companies for the future and use AI that can improve margins, for example. To implement this confidently, tactical AI systems are often tested over several months. The question is to what extent proof of success is actually possible in a specific case. This can often not only increase business value, but may also be able to spawn new business models. For example, 21strategies offers an AI technology that maps specific influencing factors on the commodity market in an algorithm and derives optimized purchasing recommendations. This enables the most advantageous time to buy the selected commodity to be calculated in advance, depending on the quantity. But the use of tactical AI goes one step further. For example, AI technology has the potential to transform a commodity trading company from a mere physical business into a more digital and hybrid one.
Tactical AI allows companies to create an additional economic driver. However, companies often overlook the fact that an application of AI systems also involves a crucial procurement and appropriate provision of data, which can also become costly. The use of AI in the enterprise relies on the availability of data. Data must be available or procurable in sufficient quantity and quality. While historic data is used to set up the contextual models of tactical AI, real-time operational data is required to make proper algorithmic decisions. Should it be missing or should the procurement become costly, the business case is also under scrutiny. Thus, a prerequisite that seems so simple can become a decisive factor.
Success factor Two for successful integration of AI systems: Processes, workflows and teams
For tactical AI projects to be successful, several criteria must be met. And this goes beyond just a business case. Concepts often find proof of success. Unfortunately, this just neglects the fact that people need to adapt to the use and application of tactical AI systems. There must be a structure in place that allows for the integration of tactical AI through flexibility and openness. Accordingly, there must be a process fit. But AI also changes existing business processes. AI can thus be disruptive. In order to successfully implement tactical AI, processes and workflows often also have to be completely rethought and redesigned to get the most from tactical AI, its speed and decision-making. Success is thus based on the openness of processes, workflows and not least of entire teams to adapt structures with newly proposed solutions.
Success factor Three for the successful integration of AI systems: The corporate philosophy
The question here is to what extent organizations are willing to relinquish control. Prof. Yvonne Hofstetter points out that companies and therefore people want to trust an AI system just as much as they trust a human being. Moreover, it is human to make mistakes. But a tactical AI is not forgiven for making mistakes. The question is, to what extent is one willing to relinquish control as a human being and as a company in principle? This is basically easy when it comes to contained systems such as the use of translation tools in daily work. When it comes to complex corporate issues, such as in the legal department or hedging capital market risks, the situation is somewhat different. Do you want to be in the loop or are you willing to be out of the loop? Is there a corporate philosophy that accepts tactical AI and is thus prepared to relinquish control and trust the AI?
Successful implementation and use of tactical AI in companies thus goes far beyond what AI can achieve. The question is how open a company and their people are. These are questions that every company should answer before implementing tactical AI. We are glad to present how to observe these success criteria and be successful in the use of tactical AI.
Author: Tanja Zimmermann