Data, data, everywhere. No, supply chain doesn't lack raw data, but high-quality data? That's a different story. Accurate, timely and comprehensive data can significantly enhance the visibility and use of emerging tech tools that enable companies to respond swiftly to market changes and disruptions.
I recently covered many AI misconceptions in the industry, but digging into data relates directly to how AI performs. A recent study found more than 45% of newly created data records had at least one critical fault preventing them from being used for training an AI model. Heather Hoover-Salomon, CEO of uShip, says logistics companies need to get off the fence when it comes to data curation for AI.
"All aspects of AI, from machine learning to predictive and descriptive analytics, are dependent on massive, curated datasets to function properly. Despite mountains of proprietary historical pricing data, AI is not yet a magic bullet for the industry since most of it isn’t properly structured," says Hoover-Salomon. "Before utilizing data sets for model training, logistics pros must first carefully check for any missing values, outliers or other abnormalities that may cause errors once the data is processed. This is why data scientists spend more time on data validation than any other step in AI development, since most of these checks happen manually. Shoring up one’s data warehouse before applying AI is a critical step in deriving business value from that data."
High-quality data is indispensable in supply chain management, serving as the foundation for operational efficiency and accurate forecasting. Hoover-Salomon explains that, thanks to decades of investment in data collection and analytics, the supply chain and logistics industry could be the poster child for the impact of data-driven AI innovation.
As emerging technologies continue to evolve and integrate into supply chain processes, the importance of robust data will only grow; Hoover-Salomon says, with 50% of logistics pros reporting a prevalent “lack of digital culture and training” in their organizations, a stronger focus industry-wide on digital literacy (including how to curate AI-compatible data) is the first step.
