Too many organizations are taking AI for granted. Too often it is assumed it will “work” as advertised by the vendor. It is assumed that AI is mature enough that if given the data it needs, it will produce results. It is also assumed that a solution to a problem that is based on AI or built around an AI-based model is needed. AI is the new silver bullet, so it surely can help with every challenge, right? Wrong. Too often, traditional technologies and techniques might be superior and ready to solve problems; however, these are often being bypassed by users who just jump on the AI-based offering. This leads to excessive costs, and worse, failed implementations and unmet challenges.
In addition, organizations tend to take a “scatter-gun” approach with AI and do proof of concepts all over the place. It would be more effective to understand the nature of the technology and its capabilities, and limitations, to target specific areas where AI can add value. Plus, the feedback analysis that helps AI learn how the predictive output can be further tuned is a critical capability often overlooked.