A BERT- the Bidirectional Encode representation from Transformation, is an AI that researchers developed. And they connected this AI to study the reviews on Amazon and differentiate between good and bad food.
The Food and Drug Administration takes a bucket load of time before verifying the bad food and issuing a product recall. So, many recalls come after people suffer and there is no use of recalls. This process isn’t fast enough to report the unsafe food.
The delay in the process and no prompt action made the Medical associations take a huge step. There was first mention of this AI in the Journal of the American Medical Informatics and Association(JAMIA). It is here that the researchers talked about BERT for the first time.
They also mentioned that the AI has a 74% accuracy. But with the integration with Facebook and Twitter, the AI can be more accurate. BERT is a smart AI that identified 20,000 reviews from a review site, which have mentioned about the uninvestigated state of the food products.
Health Departments are already taking note of the data from Twitter, Google as well as Yelp to monitor food borne diseases. Docyor Elaine Nsoesie who is the assistant professor of Global health of BUSPH mentioned how BERT is a tool that smartly uses information . She said how it uses the dat by health departments as well as the customer reviews.
It has a unique process that helps the AI work remarkably well. And It works on any reviews that have the words Sick, Label, ill, foul and rotten. It picks up reviews with such words and identifies the foul food. Even though it is not 100% accurate, it works most of the time.