An unexpected sour taste in a piece of chocolate is more than just a minor disappointment; it signifies food spoilage, an issue that artificial intelligence is now helping to mitigate. According to a recent study by researchers from Clemson University’s Department of Chemistry, published in the “Journal of Agricultural and Food Chemistry,” AI algorithms are guiding scientists toward new antioxidant blends that extend the shelf life of various food items.
The Science of Food Spoilage
Food spoilage occurs when it is exposed to air for an extended period, leading to oxidation. This process allows common ingredients, especially fats and oils, to react with oxygen. Additional factors like heat or ultraviolet light can accelerate the process. The outcome is the formation of smaller molecules such as ketones, aldehydes, and fatty acids, giving spoiled food a distinctive taste and strong metallic odor. These can pose health risks if ingested regularly.
Antioxidants serve as a protective shield against this deterioration process, both in natural and processed foods. They comprise a broad range of molecules, such as Vitamin C and various synthetic compounds capable of defending against oxidation.
The Complexity of Using Antioxidants
Utilizing antioxidants is more complicated than merely sprinkling some Vitamin C on food and expecting a protective effect. Researchers need to meticulously select and calculate the proportions of various antioxidant blends. However, mixing them does not always yield enhanced protective effects. Incorrect or disproportionate blending can result in a phenomenon known as “antagonism,” which reduces their protective capabilities.
AI in the Quest for Optimal Blends
The challenge of discovering suitable antioxidant blends for different types of food involves extensive, time-consuming, and expensive experiments. Lucas Eris and Carlos de Garcia, Professors of Chemistry at Clemson University, say that they wanted to teach AI how to identify new antioxidant blends. They trained the AI program with a database containing around one million chemical reactions and basic chemical concepts.
“Once the program could recognize general chemical patterns, we activated it by introducing some advanced chemistry,” they explained in a shared article on ‘The Conversation’. “In this phase, we utilized a database featuring about 1,100 mixtures previously described in scientific literature.”
The AI can now predict the outcome of combining any two or three antioxidants in less than a second. However, these predictions are not always fully aligned with real-world experiments. “We found that the AI was only able to accurately predict some of the oxidation experiments we conducted using real fats, indicating the complexities involved in transferring results from a computer to the lab,” the researchers concluded.
By employing artificial intelligence in this manner, scientists are not only breaking new ground in food preservation but also taking significant strides towards reducing waste and enhancing public health.
