Artificial Intelligence isn’t going to replace company CEOs any time soon. But what AI is already doing is pretty impressive: uncovering unique connections and relationships that can help businesses become more efficient at what they do, developing insight into new products and solutions, and creating a more frictionless trading landscape.
Copyright by business.financialpost.com
Adopting the advantages of AI can also present challenges, from providing customers with “explainability” for AI decision-making to ensuring that users understand both its disruptive possibilities and its limitations. That’s why developing AI to leverage Big Data has been a major focus for HSBC Global Banking and Markets (GBM). “Our access to more than 400 sources of information can help us find natural trading partners who may never have been aware of each other,” Matthew Sattler, director of business development & data science, data analytics at HSBC Bank plc (“HSBC”), GBM in London, has been working on a far-reaching plan to collect, organize and synchronize the bank’s data assets across all of its global markets, then leverage that data to create value for HSBC and its clients.
“Generative AI, where we picture machines programming themselves without a human intervention, doesn’t exist in practicality,” says Sattler. “But we are pushing the envelope of using machine learning, a subset of AI where we can derive tremendous value for our clients. For example, we can deploy deep learning models where a machine is weighing the importance of hundreds of features in large and complex data sets to build new insights through hidden patterns that are perfect for a machine to mine. With enough data, it can then make decisions similar to a person and becomes more accurate over time, using more data. The result means data becomes a strategic business asset and helps put the client first through more tailored products and services and allowing the data to speak to us in new ways.”
Sattler’s team at HSBC GBM, for example, deployed machine learning as it sought the most accurate way to create natural groupings for HSBC’s clients across the globe. The results were surprising. Instead of grouping clothing manufacturers together, for example, machine learning revealed commonalities among clients that went beyond the market segment they served.
“It’s not just about assets under management, sales turnover or market sector,” says Sattler. “Data and machine learning allowed us to see beyond those conventional tierings and find that the products and services these companies needed were different from what we might otherwise have thought. We like to call this behavioural fingerprinting, which means we are able to recommend the right product at the right time given the current client financial lifecycle. If a buyer enters a new jurisdiction to do business with a new supplier then it is important that they have the right liquidity and currency management structure in place. Our tools are able to mine for these insights and thus build a tailored product, based upon our client’s current and projected business model.”
Sattler notes that AI and machine learning depend on solid and accurate data to derive worthwhile conclusions. Bad data simply makes organizations susceptible to bad decisions. […]