The 21st Century Role of Machine Learning in the Food Industry

I was reading this article from the Food Marketing Institute about technology and was going to write a post, but as usual, my thoughts were too verbose so I had to include it in an attachment. What are your thoughts regarding machine learning and the food industry? What are your experiences? Good? Bad?

The food and beverage industry is an ever growing market, one where innovations in machine learning can enhance both food safety and quality assurance. Though this seems to be a topic of great intimidation and concern for some, lets look at some ways that it is already at work in our industry. A significant advantage to the use of artificial intelligence (AI), or machine learning is that we can learn from our past experiences through data analysis in a fashion that is far more expeditious and accurate than what our human counterparts can produce. These technologies are already affecting the way we work though the following ways:

Changing the way we work. A common sentiment when any discussion of automated processes and artificial intelligence is involved is that the robots will take all of our human jobs and we will have no where to work. This is a vastly illogical fear. There will be work, though it may be a different type of work than was completed previously. When a job is displaced due to automation, another opens due to a need for the specialized knowledge it takes to maintain the processes and equipment. Given that machine learning is still in its infancy, one can only assume that the way we work in the future will be much different than the way we do now. 

Provision of continuous monitoring in both an efficient and effective manner. Gone are the days where we had to rely upon manual temperature monitoring to ensure that our equipment is maintaining the proper environments. Continuous monitoring of temperature control can ensure that food is maintained safe and costly food losses are averted. In my career, there have many instances where $10,000 to $20,000 USD are lost when a walk in goes down after closing, then for the employees to come back after a day the business was closed to find the walk-in cooler not working. That has all changed. Now, during your day off, or during the night, you would receive a text from your refrigerator telling you that the temperatures were increasing so you can address the issue before it becomes a major loss. They can also keep a data log of temperatures that you can use to demonstrate compliance with health and safety codes and ensure that your products are not temperature abused. This is just one of the many ways that monitoring can occur. There are other techniques such as lasers, x-rays, spectroscopy, and cameras that can examine both the internal and external characteristics of a product at any point in the harvesting and manufacturing process. These technologies have advanced product sorting to a point where yields are much better than they ever were using the conventional produce sorting systems.

Improvements to sanitation using automated processes. It is imperative for food safety that the equipment used in processing are cleaned at optimized schedules. This has long been a concern when relying on manual cleaning – was the cleaning completed on schedule, was it completed properly, and so forth. The advent of clean in place (CIP) systems has revolutionized this process with their ability to be programmed to clean on predetermined, timed cycles. This further reduces risk by minimizing human contact which can reduce the potential for contamination.

Precision Traceability. Recalls happen every day, and in the past, it was quite a slow, time consuming endeavor to track down, gather, and interpret the data involved. Today, thanks to the use of artificial intelligence, we are able to compare historical data and predict certain events across multiple periods of time from different areas in a way that allows us to report and employ strategic safety interventions rapidly to protect our consumers. An example of how this works at present is through how Amazon uses AI to predict recalls based on feedback they receive from their consumers. This process is so accurate that it is able to predict a recall and allow Amazon to block it’s sale an average of 50 days before the official recall occurs. Frankly, that’s astounding.

A final word. There are so many ways that AI and machine learning can make our food supply safer that we cannot yet even begin to fathom. Though there is much knowledge about machine learning, it is an evolving science, so we should not erroneously believe that we completely understand its capabilities. I believe that there will be things happening in this realm in the next twenty years that we could never even imagine at this juncture, therefore it is prudent that we maintain an open mind about how these changes can affect the way we live, work, and eat.