As the buzz around Artificial Intelligence (AI) continues to gather momentum, Philippe Laurensy, CEO of Asia Pacific at Euroclear, spoke at The Network Forum (TNF) conference about how the technology is transforming post-trade.
If AI evangelists are to be believed, the technology will help the securities services industry unlock hidden value and maximise productivity, thereby generating efficiencies and potentially countering some of the cost and revenue challenges it is facing elsewhere.
Rather than being homogenous in nature, AI models are incredibly diverse.
Machine learning tools, i.e. Optical Character Recognition (OCR), Natural Language Processing (NLP), etc, have existed for years now and are deployed widely across a number of post-trade processes.
"At Euroclear, we have been using AI for around 10 years in the form of machine learning and Robotic Process Automation (RPA) to streamline inefficient processes", according to Laurensy.
For instance, RPA has been utilised by post-trade providers to expedite client and regulatory reporting, allowing for people to spend more time on value-add activities.
At Euroclear, AI is also fairly ubiquitous in sanctions screening. "AI tools can go through all of the various sanctions rules, allowing us to identify or flag issues much faster," said Laurensy.
More recently, Euroclear has been leveraging predictive analytics to help clients manage settlement risk. By scouring through large quantities of settlement data, custodians and Central Securities Depositories (CSDs) can spot potential fail trends and notify clients accordingly, so they can take corrective measures.
"By intervening quickly and addressing possible settlement fails, clients can avoid penalties for trade fails under the Central Securities Depositories Regulation (CSDR). Additionally, the transition to T+1 settlement cycles will also become easier for them if they improve their settlement discipline," said Laurensy.
"At Euroclear, we have been using AI for around 10 years in the form of machine learning and Robotic Process Automation (RPA) to streamline inefficient processes."
Philippe Laurensy - CEO of Asia Pacific, Euroclear
The debate about AI has been turbocharged following the emergence of Generative AI and Large Language Models (LLMs), such as ChatGPT.
These tools are trained on vast quantities of online written and visual content and are capable of producing human-like text and imagery, which many believe will lead to widespread productivity benefits across multiple industries.
According to the McKinsey Global Institute, the increased productivity enabled through Generative AI could add anywhere between USD200 billion and USD340 billion in value annually to the global banking sector, corresponding to roughly between 2.8% - 4.7% of total industry revenues. (www.mckinsey.com)
So, how is Generative AI being used in securities services?
"At the most basic level, a lot of people in our industry are using tools like Microsoft Co-Pilot to provide summaries of in person team meetings or to create abridged versions of large documents. This is helping them to save a lot of time in their day to day lives," said Laurensy.
Although Chatbots have long been synonymous with poor customer service, many post-trade providers are now starting to integrate Generative AI into Chatbots and the results have so far been impressive.
During TNF, a number of experts highlighted that agent banks are looking to use Generative AI-enabled Chatbots to answer not just rudimentary client questions (i.e. information about trade status updates) but increasingly complex inquiries, such as how a niche local regulation could impact market access or trading.
In network management, Generative AI could help speed up operational due diligences on agent banks, said Laurensy. "The Association for Financial Markets in Europe’s (AFME) Due Diligence Questionnaire (DDQ), which network managers send out to agent banks as part of their due diligences, has roughly 600 questions, of which circa 590 are fairly harmonised. There is no reason why Generative AI tools could not be used to answer these broader questions, allowing network managers to focus their attention on more pressing issues," said Laurensy.
As with any nascent technology, the potential risk impacts of AI are still yet to be properly understood.
Perhaps the biggest challenge facing AI is its data.
If an AI model is built or trained on defective data, then the analytics it churns out will be flawed. Should a provider share that bad data with a client, then this could lead to operational risks or even losses.
"ChatGPT and Generative AI assume that everything on the internet is truth and the technology does not understand what it is writing. This is something people need to be aware of, especially if they are incorporating this technology into their decision-making processes," said Laurensy.
People’s long-term cognitive development also risks being blunted if they end up becoming too reliant on AI in their everyday lives.
"If a person is using AI constantly, then they will not be engaging their brain. If you lose your capacity to do certain skills, then this could have profound implications on our industry and beyond," added Laurensy.
Some regulators have also started introducing AI legislation, which firms will need to keep a close eye on.
Although the US appears to be taking a lighter touch approach to regulating AI, the EU is widely considered to be ahead of the curve following the introduction of its AI Act.
The EU’s rules prescribe tough disclosure requirements on firms developing what regulators describe as being high-risk AI systems, although general purpose AI models – such as those used in post-trade – will be subject to less scrutiny. (www.reuters.com – 22 May 2024)
AI is going to have a significant impact on the industry’s operating model, by driving up productivity and facilitating massive operational savings.
At the same time, the risks of AI (i.e. bad data risk, etc.) cannot be ignored.
One of the biggest issues facing firms, however, is making sure that the efficiency gains enabled by AI does not come at the expense of people forgetting vital life and career skills.
