How Accenture is leveraging LLM’s to help enterprises with product discovery

Accenture has revealed its R&D team is in the midst of testing large language models (LLM’s) such as generative AI in the food industry to boost knowledge discovery and augment creativity.

The Dublin-based IT and professional services provider gave a virtual tour of one of its seven innovation labs in Sophia Antipolis, France. During the tour, speakers from the IT company gave the example of baby food, in particular spaghetti bolognese for a 12-month old.

It first needed to gather masses of information from social media and feedback from the consumers before applying AI as part of the knowledge discovery process.

The AI, which has been built into Accenture’s interface, listed popular ingredients, competitors who make that particular dish, plus positives and negatives for customers. This is all pretty typical in knowledge discovery, according to lab director Anne Groeppelin who led the virtual tour.

“What the team are now doing is looking at the ingredients in that formula, looking at all the criteria in terms of nutrition, the consumer rating, the cost, where the ingredients have been sourced in that recipe, the sustainability of the product and its taste profile,” she said.

The R&D team have developed a tool which allows its clients to change the recipe using suggested ingredients from generative AI. These are listed into three categories: expected, surprising and novel.

“The full idea was to add that ingredient [cocoa] which changes the formulation, the nutritional information and the cost, and then I can make the decision, do I want to try this formulation or not?” asked Groeppelin.

According to Groeppelin the tool not only boosts creativity but also efficiency, using simulation through generative AI to test what would be the best formulation for the recipe.

“If I add cocoa to that recipe, what is the amount I should add to maintain the nutrients score, to maintain the costs at an acceptable level, or to maintain the sustainability?

“It’s an example where we are using AI to augment creativity to interpolate new ideas and use large language model types of AI to reformulate and get to the level of detail that is needed to actually produce that product.”

Accenture is also testing LLM’s in the pharmaceutical industry to reduce the time to produce and market medicines and vaccines.

The lab director said it takes, on average, ten years to develop a new drug and get it approved due to the complexities involved in knowledge discovery.

“In life science in particular, you have very complex information from the existing drugs, the genes that are sources of a particular disease, regulations, tests of the medicines with the patients, and this amount of information cannot be managed by even several brains together.”

Accenture’s innovation lab in Dublin is leveraging LLM’s to perform knowledge discovery across a large set of information, linking all the existing information to predict the link between diseases, drugs, treatments and medicines.

For example, take Alzheimer, what are the types of medicine that has been applied before and what are the types of scientific publications that relate to it? Groeppelin asked the virtual audience.

“It’s really plugging all the information together to detect information we may not yet know and unearth new insights.”

AI is also being tested in drug product discovery to look at a specific molecule and simulate an alternative one that could have a better effect on the treatment of a disease.

“We select a molecule and decide the protein we want to attack in the body and then it interpolates what other molecule could fit, and those are already ones that have been approved by the FDA [Food and Drug Administration].”