The point of this is not to teach every DBS employee to be a machine-learning expert. It is to help them to communicate with those in the business whose whole world is data. It allows staff to understand what machine learning is about and the basic principles of how it works. It puts them all somewhat on the same page.
“We have training for senior leaders and for our analytics resources, but what we have seen is that very often these two groups of people speak different languages,” says Jurgen Meerschaege, head of data culture and curriculum in DBS’s transformation group.
“Businesspeople are aware of the opportunities that are available, but are not aware of the capabilities of AI and machine learning. On the other hand, you have this crew of technically and mathematically able people who don’t necessarily share a common interest with the bankers, but just want to solve problems with complex models. To bridge that, you need a translator.”
A translator, in this sense, is somebody who understands enough about business and enough about machine learning to bridge both worlds, to ask the right questions of data scientists and to understand how they might best be applied in banking.
McKinsey has said that a truly data-driven organization needs 10% to 15% of the workforce to have these translator skills.
“We have never believed in limiting digital expertise to a small team,” Paul Cobban, chief data and transformation officer at DBS, said last month.
DBS was not the first bank to use this system; when its bankers first started using it, getting good enough to qualify for national annual championships, they found the leaderboard full of staff from Capital One.
However, DBS has rolled it out comprehensively as a form of training, launching its own DBS x AWS DeepRacer League, and by the end of the year 3,000 of its staff will have undergone the programme, including the bank’s senior leadership.
Addictive
They, too, can find it an addictive process. To excel at controlling the car, you get into a host of other considerations – for example, taking the car selectively off the centre to pick the racing line or teaching it how to drift. The standard worldwide is highly competitive, culminating in annual championships usually held in Las Vegas – though done virtually for the foreseeable future.
DBS staff now account for about 10% of the 90 or so qualifiers who get into that championship. One of them, my DBS trainer and executive director of technology and operations Ray Goh, won in May 2020, beating Formula 1 driver Daniel Ricciardo and his team on a mocked-up track of the Barcelona-Catalunya circuit.
The immediate banking parallels to this whole process would be areas such as the more sophisticated chatbot functions. DBS has an internal one, called HiRi, for HR-related queries; also a chat function in the bank’s SME business, called Joy. Simplified end-to-end credit-processing techniques also use machine learning to help with credit risk management.
However, the idea is not for everyone who takes this course to be able to devise a breakthrough in machine learning. It is to enable them to talk to each other in something approaching the same language.
Read the whole article here
