At the 25th Annual ISITC Securities Operations Summit, in Boston, I joined a panel of industry experts to discuss deployment of machine learning, artificial intelligence (AI), and robotics, as our industry looks to intelligent automation to accelerate workflows, increase efficiencies and reveal new revenue streams.
Joining me on the panel, “Why AI and Robotics: The Allure of Intelligent Automation,” was Jason Baldesare, Director, Strategy and Solutions Management FIS; Dushyanth Sekhar, Senior Director, Center for Automation, Robotics & Extraction, S&P Global; and Laurie Stiles, Global Head of Data Management, Bloomberg.
During our discussion, we covered topics ranging from business strategy to IT investment, risk analysis and the value to be created using these emerging technologies. In fact, DTCC has published white papers, held cross-departmental working groups and embarked on a series of tests to identify (and then prove) the business case for automation within different scenarios, and we have begun using these technologies in many parts of the organization with positive business outcomes - as have my co-panelists in their businesses as well.
However, within the industry, it’s becoming clear that there’s a real risk of taking emerging technologies and misapplying them to business cases. Much of this comes from not grasping the benefits, and challenges, of AI and automation - both of which cover a large range of individual protocols, programs and use cases.
Most importantly, we want to avoid using “tech for tech’s sake” - it’s got to make sense for the business, and be grounded in a real understanding of what our clients need. With these technologies, as with any technology, our focus is delivering value to our clients and addressing their business needs. This is true of all technologies. But with RPA and AI, we can get caught up with the hype and too often can risk putting the technology first, ahead of the business opportunity or problem.
At DTCC, we take a measured approach when using these advanced tools and start with the business requirements first - not the technology. We ask ourselves: What problem are we trying to solve? Once we’ve identified that, we then look at the technologies available that could help us solve it.
Definitions: AI vs Automation
As a quick definition, automation, including robotics, can be described as technology that mimics the tasks a person does with their hands on a computer (e.g., data entry, moving documents around, accessing web apps, and so on) and is a highly optimized structured set of processes and tasks.
AI, on the other hand, seeks to mimic the tasks a person does with their head: decisions, analysis, forward predictions and pattern-spotting. In some cases, there may not be any actions per se that have to be taken. Instead, we might use AI to ingest a set of documents and parse and classify the information, before presenting it back to the user with useful insights.
Getting Started
During the summit’s Q&A segment, several people asked how to get started with these emerging fintech tools. We all shared different use cases, but the simple rule of thumb is one we all agreed on. You can truly understand a new technology only by using it. Meeting with advisers and consultants is always helpful. But taking the time to use the technologies – hands on experience – is key to grasping the potential and real-world applications. It's also a good idea to test the technologies in a sandbox.
For example, in our Mutual Fund Services, we expanded and automated our Security Issue Database by extracted from text-based documents (which is quite a different thing than using tabular and numerical documents like spreadsheets or databases). The focus for the service is the large repository of Mutual Fund prospectuses and other documents filed with the SEC’s EDGAR system. There’s a vast amount of data contained there and manual data analysis isn’t time efficient. Having a large group of people read the documents would be too slow, costly and potentially inconsistent and prone to error.
Instead, we deployed AI to gather, read, catalog, organize, analyze and present everything into a web-based, client-facing dashboard which creates huge value for our clients. That’s a good example of using advanced fintech - because it makes sense for the business, the client and a suitable scenario for AI’s skill set.
To this end, it’s important to note that we’ve placed some – but not all - of these new technologies into the departments responsible for the business rather than within IT. For example, we have a robotics automation team, as well as several distinct AI and data science units, embedded within non-IT divisions, and managed directly there.
This has helped us because when the business staff have oversight and responsibility for emerging technologies, they have the opportunity to “own it” and be the first to determine whether or not it’s working. We can then use those learnings for other business lines to scale robustly.
Robots vs Humans
At the ISITC summit, the issue of bots replacing humans came up, as could be expected. This question has been in the news recently, and not for the first time – at least partly in response to the renewed interest in robotics, AI and automation. Unfortunately, it’s been blown out of all proportion by the media with scary stories around future “Skynet” Terminator-style systems.
Once we put the business opportunities and problems first, this highly controversial subject tends to go away. Our purpose in using robotics and AI is not to eliminate jobs: it's about extending the ability to execute work, having staff focus on higher value (and more interesting) tasks, reducing risk and scaling the business overall.
Most directly, automation is a way to reduce the risks and inefficiencies that come with manual effort. Human minds can wander, which can lead to mistakes. Bots, though, don’t get distracted and don’t get bored or tired. Bots are reliable - with proper oversight and well-designed programming - and more suited to certain mind-numbing tasks.
Future Tests: Behavioral Analysis
Looking to the future, we’ve been exploring the use of AI to provide sentiment and behavioral analysis. The use case for this is interesting: as our clients start to deploy robotics to access DTCC applications, we can use AI to help detect a range of high-risk and threat behaviors, such as phishing or other malicious attacks. We can also use AI to distinguish legitimate bots used by our clients from malicious bots used by hackers to probe and attack our systems. This requires an advanced level of linguistic and other usage modeling which is more difficult to do with traditional technologies. AI now enables us to distinguish between human end users and robotic systems, crucial when protecting our systems against cyber-security threats.
Essentially, our applications were designed to be used by people, not bots, and so we have to work with our clients to roll out guidelines to keep robotics technology use within safe boundaries. In tandem, we also want to ensure firms can facilitate their use of robotics, so we’re designing, developing and testing our own future systems to be responsive to the end user.
There is tremendous interest in the use of robotics, AI and intelligent automation to advance not only individual competitive advantage but also our industry as a whole. While we are just at the beginning of a long journey into these advanced technologies, it is already clear that their promise is real, and the opportunities to reduce risk and cost, and to promote new ways of working and transacting, are here today.