The rush to build artificial intelligence applications has hit a wall of poor quality data and data complexity that’s hindering them from taking advantage of the technology. Those barriers are also preventing firms from upgrading other parts of their tech stacks.
A slew of surveys and comments by researchers and vendors paint a picture of an industry struggling to make the most of AI’s promise, with reconciliation processes, analytics and alternative asset investment also hampered by data shortcomings.More than two years since ChatGPT made headlines of generative AI (GenAI) and sparked a clamour among financial institutions to adopt the large language models on which the technology runs, organisations are realising the crucial role of data in successful adoption of these capabilities. But as a new era in economic history unfolds, with old assumptions about growth, trade and geopolitics spawning new investment and risk theses, organisations will need to fall more heavily on data to get them through the turmoil.
Without a strong data foundation, organisations will struggle to survive.
GenAI Spending
FinTech giant Broadridge’s fifth annual Digital Transformation and Next-Gen Technology Study established that while institutions are increasing their technology spend to boost their AI and other capabilities, they continue to experience fundamental data challenges.Anecdotal and qualitative evidence from more than 500 technology and operations leaders across wealth management, capital markets and asset management firms globally, found that 80 per cent were planning moderate-to-large investments in AI this year. Researcher Gartner estimates that total global spend on GenAI, in particular, would be US$644 billion this year.
Nevertheless, their technology investment plans are being stifled by fragmented tech stacks as well as data consistency and quality.
Almost half the respondents to Broadridge’s survey said they were still battling data silos in their enterprises and two-fifths said that they had data quality issues. Nevertheless two-thirds recognised that a clear data strategy would enable them to maximise their return on investment.
“Just because you’ve put all your data in a big lake, it doesn’t mean we can all go swimming,” Broadridge head of international strategy and corporate development Stephanie Clarke said in the report. “The real value lies in enabling disparate data sets to communicate effectively with one another, guided by a well-defined data and governance strategy.”
Trust in Data
The Broadridge report also suggested that market participants were still wary of trusting AI, which a Gartner study and discussion said was being amplified by poor data quality.
“Trusted, high-quality data is key to enabling a data-driven enterprise,” Gartner wrote in a report that accompanies a discussion tour of the world that the company has launched. “Trust models look at the value and risk of data and provide a trust rating based on lineage and curation.”
It’s not just the implementation of AI applications that are being hobbled by poor data estates. According to a survey of asset management and capital markets participants by data management services provider AutoRek, inadequate technology stacks continue to contribute to the data challenges that are creating inefficiencies for raft of critical processes.
The company’s report, based on 250 interviews with senior managers in the US, found that four-fifths struggled with reconciliation due to the overwhelming volume of data that entered their systems, the level of manual processing required and difficulty in data matching.
They said the three challenges that were most difficult to overcome were data integration, regulatory reporting and the surge in transaction volumes. At the same time, almost two-thirds of respondents said they relied on spreadsheets or legacy systems to perform their reconciliation activities, with a fifth saying they used only spreadsheets. More than half said manual processes were a “significant contributor to workload challenges”.
‘Worrying Picture’
The report concluded that a lack of investment had meant too many organisations were reliant on outdated processes that create inefficiencies and make it even more difficult to map large datasets.
“A worrying picture has developed that shows how decades of underinvestment in back-office operations has created inefficiencies that undermine performance, compliance and scalability,” wrote Jack Niven, Vice President of North America at AutoRek.
“While there is recognition from the firms we surveyed that automation and evolving technologies would be beneficial, years of inaction have created an alarming prospect of widespread operational disaster.”
Subscribe to our newsletter
