AI systems, while capable of tasks like writing poetry and summarizing texts, often struggle with basic arithmetic. This limitation arises because AI is designed to identify patterns in data rather than perform precise calculations. For instance, the latest version of ChatGPT achieved only 64% accuracy on math problems, which is concerning for business applications where exact numbers are crucial.

Most large organizations deal with messy and inconsistent data due to years of quick fixes, acquisitions, and neglect. This 'technical debt' complicates AI application, as AI requires clean, reliable data to function effectively. Experts suggest that until companies address these data issues, the potential benefits of AI will remain unrealized.

Siloed expertise can significantly hinder AI initiatives. Often, those who understand AI do not fully grasp the business problems that need solving, leading to a lack of collaboration. Successful AI implementation requires a partnership between data scientists and business staff to ensure that AI solutions are effectively integrated into operations. Without this collaboration, investments in AI may not yield the expected results.