While autonomous AI is a feasible possibility, the industry “shouldn't expect it anytime soon”, a speaker at this year’s Sibos conference in Toronto predicted.
The comment was made by Kevin Blair, senior vice president of asset servicing at Northern Trust, during the ‘Closing the gap to full automation in asset servicing’ discussion.
Blair began the session by outlining the external factors driving the industry’s growing interest in automation. Accelerated settlements, regulatory increases, increased volumes and fee pressures are all encouraging greater automation, he said, as are changing demands from a new generation of market participants. Data is now expected to be transparent, accessible and delivered on a real-time basis, which automation can assist with.
Touching on the collaborative finance theme of the conference, Blair highlighted the importance of working as an industry to ensure that there is standardisation across the board when it comes to evolving technologies. Northern Trust is working with vendors to further automate routine processes in order to free up more time for activities that add value for clients, he explained, with strategic partnerships and investments helping to create products that can benefit participants across the industry.
Another key topic at this year’s Sibos is data, which has a key role to play in the development of automated processes. Blair emphasised the importance of data governance and data quality, advocating for data capture to occur earlier on in the process. This will prevent bad data “infecting” systems further down the line, he explained.
Data from external sources can also be an issue, with Blair stressing the need for industry-wide collaboration and standardisation.
Considering AI, Blair divided the wide-reaching technology into two buckets; tools mimicking human actions, which are already widely used, and tools mimicking human behaviour. The latter can learn to detect patterns and operate more complex algorithms, with use cases including anti-fraud initiatives and data extraction. Blair added that these tools are already in operation, with a large number made available on an open-source basis and able to be deployed instantly.
There are a number of challenges when it comes to implementing AI technology, with data quality, availability and bias potentially leading to suspicious or incorrect results. Blair added that firms must be able to explain the results that their AI programmes are producing, avoiding a “black box” situation.
Considering further use cases for automation in the asset servicing industry, Blair named reconciliation, file processing and predictive analytics as areas that could benefit. The latter can enable firms to optimise their working conditions and staff organisation, he explained, and provide evidence to “hunches” that would otherwise go unevidenced. However, he added that autonomous technology could prove challenging to use given the highly regulated nature of the industry.
Closing the session, moderator Sanjay Prasad, head of capital markets and wealth for the US, Canada, UK and Ireland at TCS Financial Solutions (TCS BaNCS), summarised that current real time technologies are in-themselves achieving two 9s level of efficiency and data quality is perhaps biggest road block toward the last mile. Therefore, if emerging AI suite of technology can be used to “fine-tune” and cleanse data, “taking out the noise”, it can help asset servicing firms to improve their operations from two 9s to four 9s and beyond. Blair concluded that while he believes automation has the potential to streamline processes and “make life easier”, humans will remain involved in the process for the foreseeable future.
Image: peshkova/stock.adobe.com 