How AI Tools Are the Future of Food Quality Management
September 18, 2023
Artificial intelligence (AI) is the next frontier in our work to ensure that consumers are getting the highest quality fresh products in grocery stores and restaurants. You can find AI’s problem-solving fingerprints all over your daily life, whether you’re shopping online, looking for directions, using social media or playing a video game, to name just a few.
But AI, and specifically the machine learning (ML) trained on big data, will increasingly play an important role in food quality management.
In fact, AI and ML will one day make it possible to predict the quality of a food product before it arrives at the distribution center, potentially long before you even order it, by using data associated with the product’s journey through the supply chain.
iFoodDS has released a white paper entitled “Predictive Analytics in the Food Supply Chain” (PDF) that describes how our engineers used ML and AI in the lab to achieve just that—shifting quality analytics from reactive to predictive.
This is a vision you can make a reality in your own operations, and now is the time to get ready.
Consumer surveys increasingly indicate how much people value quality in the choices they make when shopping for groceries and deciding where to dine. A March 2022 report by FMI, the Food Industry Association, revealed that shoppers put the appearance and quality of fresh produce on par with price. The TouchBistro 2022 Diner Trends Report released in September 2022 found that 68% of 2,600 diners surveyed rank food quality as the most important factor.
Growing, delivering, and selling affordable, high-quality perishables is critical to being a leader in the industry. iFoodDS has long anticipated that AI will have an important role in creating a safer food supply, co-authoring a 2020 article in Food Protection Trends (PDF) that called on retailers to be open to using AI tools that provide actionable data throughout the supply chain to manage food safety risks.
So, let’s look at where we are now and how we can move forward.
The Modern Food Quality Inspection Process
While this process is evolving, many distribution centers still use pencil and paper to record the results of quality inspections, which leads to inconsistent inspection practices. Such inconsistencies can lead to an unpredictable consumer experience in stores and restaurants, which affects sales and customer loyalty.
Paper inspection processes also don’t allow for tracking the quality of perishables provided by suppliers over time, limiting visibility into their historical performance during different seasons, weather cycles, and more.
Not everyone uses paper. Other distribution centers use digital tools to improve inspectors’ effectiveness. More advanced operations use the data they collect during the inspection process to optimize sourcing decisions.
Solution providers have also been helping to advance this process. For example, iFoodDS offers its Quality Insights solution that uses inspection metrics to provide important information about which suppliers, commodities and locations perform the best—or worst.
However, instead of only using quality inspection data to report on the historical performance of suppliers and commodities, imagine if you could use that same data to predict the quality of an incoming product.
In the white paper drafted by iFoodDS (PDF link), we outline the data preparation process, feature engineering, model development, and model evaluation to show how brands could predict quality based on historical inspection results. In this white paper specifically, we predict the unsatisfactory rates of produce at a distribution center based on previous orders received from suppliers.
There is tremendous untapped potential in how we use data to achieve operational consistency and efficiency, and gain insights into supplier performance. The data can also be used to strengthen supplier partnerships. The result will be better selection, higher quality, and fresher product.
Using AI and ML in Food Quality Inspections
AI and ML advancements will usher in a shift from reactive to proactive strategizing. However, there is often confusion about the difference between AI and ML.
Broadly speaking, AI uses computers to process large amounts of data in ways that people cannot. ML is the use of computer systems that can learn and adapt by using algorithms and statistical models to analyze patterns in data.
Data analytics technologies utilize past historical information and advance based on new data, so they get smarter and evolve over time to make even better quality and pricing recommendations with even more accurately forecasted impacts.
Learning from past patterns to predict quality shifts sourcing strategies from reactive to proactive—reducing waste, improving the quality of perishables, and increasing customer trust and sales. Here are some examples of how AI and ML can be used to improve quality and freshness:
- Knowing when to switch production regions based on what you’re learning about the quality and price of each commodity in different regions throughout the year.
- Improving sourcing by predicting future quality based on a supplier’s historical quality records.
- Getting ahead of the competition on market fluctuations with predictive analytics providing rapid insights into future events.
- Adjusting to adverse weather events with information on yield and product availability that will guide your decisions on issues like merchandising, specification requirements and sourcing locations.
- Optimizing shelf life and reducing food waste by helping make predictions about total shelf life and how many days a commodity has left in its shelf life by the time it reaches store shelves. Retailers can use this information to inform their stock rotation methods, ensuring that consumers have access to products at their peak freshness.
How to Start Utilizing AI and ML in Food Quality Management
This is where you come in. You may work with a solution provider to use AI and ML most effectively in your operation but the work to get there begins with you.
AI and ML need high-quality input data to learn from. So, where do these data sets come from?
The data required by the FDA’s Food Traceability Rule, also known as FSMA 204, is a great place to start. The Key Data Elements (KDEs) associated with Critical Tracking Events (CTEs) contain a wealth of information on your product’s journey from source to consumer. The rule is the information backbone that other product data can be attached to. For example, cold chain data, time in transit data, and quality data can be attached to the traceability lot code.
Ultimately this data will create a more transparent and traceable food system, but that’s not all. It can be used to predict freshness, quality, and more in the future.
FDA leaders have repeatedly said that better food safety begins with better data. AI and ML will be instrumental in converting large volumes of data into powerful, predictive information.
That said, it’s worth noting that you’ll still be in the driver’s seat—you’re not handing the wheel to a computer. AI tools will support the decisions you make to reduce risks and protect your customers, but those decisions are still yours to make.