- Hershey has been using artificial intelligence and machine learning to reduce waste in the manufacturing process for Twizzlers, according to CNBC.
- The manufacture of Twizzlers previously resulted in slight variances in the weight of the licorice sticks. The company would either have to overproduce the candy by weight to account for the variance or manually weigh the candy every 15 minutes and then adjust the machine 12 times per day to minimize waste.
- By using Microsoft's Azure cloud to track over 60 million data points in the manufacturing process, Hershey was able to pinpoint what exactly impacts the weight of the candy. The machine then learned to adjust itself throughout the manufacturing process, ultimately reducing variance in the candy's weight by 50%.
The promise of artificial intelligence technologies is starting to bear fruit in a number of industries, including food and beverage. Hershey is an early adopter of the technology, using it to help make the perfect Twizzler. Not only will the company save money as a result, but it is also creating buzz in the press for its product.
The opportunity to leverage artificial intelligence could be huge for food manufacturers. Like other companies, Hershey's products promise a certain specified weight — and they can't underdeliver on that promise. Even if it's off be just a fraction, Hershey and other manufacturers often have to overproduce to make up for any variances in weight. That adds up to a lot of overproduction — and money — over time.
Other use cases are starting to appear in the food industry as well. Wal-Mart recently filed a patent for a technology that would track consumers’ usage of everything from toothpaste to shoes, automatically placing reorders and suggesting additional products. Such a system would rely on sensors placed on products and come as a close competitor to Amazon’s Dash Buttons. If consumers ultimately adopt this kind of technology, systems like these could go a long way toward automating the shopping list. While it's still far off, the potential end-game of such a system is that household essentials like diapers, laundry detergent, coffee and other consumables would no longer need to be actively reordered.