The History of Manufacturer Recalls
As the Fourth Industrial Revolution, involving artificial intelligence and big data, is upon us, so are several product recalls. For years manufactures have promptly dealt with product contamination and malfunction in a reactive manner, sometimes leaving a wake of destruction behind. From product recalls to FDA food recalls, there have been several severe recalls that have damaged lives and company reputations semi-permanently; Infantinos Baby Slings (2010), Peanut Corp. of American in Georgia (2008), Johnson and Johnson Tylenol poisoning (1982) and Galaxy Note 7 explosions (2016). Earlier this year Ritz Crackers and Goldfish were supplied whey powder by the Associated Milk Producers Inc. (AMPI) that cause additional chaos. The latest this week? The lettuce recall, romaine lettuce, and 100,000 pounds of ground beef.
So this poses the question, why haven’t we figured out how to proactively protect consumers from the danger of food recalls? Well, technically, we have, but it isn’t ready yet. The Fourth Industrial Revolution— big data, artificial intelligence, machine learning and predictive analytics—have allowed manufacturers to gain more insight than ever into machine processes, but they are still working out the kinks.
As with any industry, the manufacturing market generates massive amounts of big data. The industry has started transitioning to smart factories, where machine learning and artificial intelligence are used in every process to predict machine failure, product defects, missing insights and much more.
While the idea of big data and predictive analytics aren’t new in the manufacturing world, perfecting, growing and evolving this type of machine learning will continue to take time. Machine learning relies on what the machine knows, it is that simple. Today, manufactures are running into one of two issues; the unknown and the expensive.
Tiny, rare problems, that lead to large recalls, can be easily missed because the system hasn’t yet learned the variance. I will say it again for the people in the back, machine learning is only good as its tested sample size. This is a problem in manufacturing market because there can be several unknown factors that lead to a recall.
On the other end of the spectrum, you have highly sensitive models that are taught to detect any issue, however this generates a substantial number of false positives and lots and lots of money. A disruption from a highly sensitive model can slow down production, decrease efficiency and obviously increase cost. The solution is discovering the best algorithm to teach the machines a happy balance of detecting known and unknown variances.
So where does that leave consumers now?
In a more elevated place than we were 20 years ago. People reports that the Center of Disease Control has investigated 22 recall outbreaks in 2018 so far. This is the highest number in 12 years. Food recalls in 2018 seem like a huge problem, however some would disagree. The number of food recalls and outbreaks show that machine learning, big data and the fourth industrial revolution are improving and evolving. The elevated technology is allowing manufactures to connect illness to a product much sooner. The manufacturing industry is moving forward one step at a time to ensure proactive safety for consumers against food recalls. Soon enough our safety assurance will be in the hands of machines entirely.
Headquartered in Houston, Texas, ethosIQ’s cloud and premised-based software has delivered business intelligence to multinational corporations and government agencies since 2009. ethosIQ’s award-winning software collects, correlates and presents data from multiple, disparate systems, empowering organizations to make informed, real-time decisions. ethosIQ software solutions provide analysis and actionable insights that enable enterprise and government organizations to deliver better customer experience while ensuring operational efficiencies and maximizing technology investments. We provide the data that enables decisions in minutes, not days or weeks.