BBVA is looking to spot and train the best in-house talent in the use of the large language models (LLMs) that are behind the success of the likes of ChatGPT, as it drives the exploration of projects based on generative artificial intelligence. To succeed in this task, it recently organized a DataRally, a competition in which some 440 employees tested their ability in interacting with one of these models. Impressed by the positive feedback received from participants, the bank may now apply the DataRally concept to the generative AI courses it will be running as part of its Data University in-house training program.
BBVA has launched several exploratory projects in recent months to gauge the potential of generative artificial intelligence in making organizational processes more efficient and generating additional added value for its customers, all the while ensuring the utmost security.
To unlock these emerging opportunities, it is essential to spot and train the best in-house talent in the use of the language models behind this technology. To achieve this, the Advanced Analytics discipline and BBVA’s GenAI team have just organized a DataRally competition. Around 440 employees from all the countries in which the bank operates put their skills to the test by taking on 12 large language model (LLM) challenges spread across various blocks. During these trials, they were tasked with processing quantitative information, finding anomalies in financial reports or identifying, scoring and classifying the level of satisfaction (or dissatisfaction) of a fictitious customer based on their comments, among other tasks. At the end of each checkpoint, the participants received feedback on what kind of prompts or commands they should have entered in order to successfully complete the challenge.
Following a grueling four weeks, Carlos Gutiérrez, data specialist at the Global Talent & Culture area, Joel de la Cruz, data scientist of the Advanced Analytics Engineering team at BBVA Spain, and Hans Hidalgo, also a data scientist of Client Solutions at BBVA Peru, posted the highest scores by overcoming all the challenges with the fewest prompts and in the shortest time.
“We couldn’t be happier with the response to the competition, which has been enthusiastic and led to healthy levels of competition,” remarked Josep Amorós, Data manager at BBVA and organizer of the DataRally. “75 percent of the participants were very satisfied with the experience and more than 95 percent would do it again. Many of them were also hugely appreciative of the bank’s efforts to offer them activities and training in Generative AI.”
“Generative AI is here to stay,” said Curro Maturana, Global Head of GenAI at BBVA, “and it has enormous potential for businesses as they revolutionize their internal processes and embrace global development projects that will allow them to deploy their products and services simultaneously in different countries. That is why we are already training the best talent at BBVA.”
To boost training in generative AI skills, BBVA will be making generative AI courses part of its in-house Data University data training program. The DataRally has been useful in testing the waters to a prompt engineering learning dynamic, which its organizers may now offer to other employees through these courses.