5
Data is the big challenge

Poor quality, fragmented data is AI’s highest hurdle, and banks need both people and tools to overcome it.

Data is the big challenge

Asked about the top challenges of scaling AI, most banks initially provide a one-word answer: data. Data is siloed in hundreds of impenetrable systems, and it’s often incomplete, low quality, in different formats or unreliable.

“The big opportunity is data quality and availability,” says Niranjan Vivekanandan, EVP and Chief Operating Officer, RBC Commercial Banking. “AI systems require huge amounts of high-quality, diverse and relevant data to learn and make accurate predictions. In a bank, we process billions of different transactions and events every single day, capturing this information across a wide range of systems. So, the effort involved in data collection, cleaning and processing is often significantly underestimated.”

Another issue is data privacy and protection. “It can be complex and time-consuming to understand whether publicly available information, such as news articles or reports from ratings agencies, can feed LLMs [large language models],” says Patrick Klasen, Managing Director, Head of Corporate Lending Process and Digital Transformation, Commerzbank.

The big opportunity is data quality and availability. AI systems require huge amounts of high-quality, diverse and relevant data to learn and make accurate predictions. In a bank, we process billions of different transactions and events every single day, capturing this information across a wide range of systems. So, the effort involved in data collection, cleaning and processing is often significantly underestimated.

Niranjan Vivekanandan, EVP and Chief Operating Officer, Commercial Banking, RBC

We’re looking for AI tools to help with data quality and do things like match and merge, automate metadata extraction and do quality scoring, but they are just not mature enough.

Cindy Saw, Head of IT Group, Data and Analytics, BDO Unibank

It can be complex and time consuming to understand whether publicly available information, such as news articles or reports from ratings agencies, can feed LLMs [large language models].

Patrick Klasen, Managing Director, Head of Corporate Lending Process and Digital Transformation, Commerzbank

Could AI make data better?

The standard approach to this challenge is to devote human resources to it. But this has its limits and is expensive, so many banks are exploring AI-powered tools that help address data quality.

“We’re looking for AI tools to help with data quality and do things like match and merge, automate metadata extraction and do quality scoring, but they are just not mature enough.” says Cindy Saw, Head of IT Group, Data and Analytics at BDO Unibank.

“We used to say garbage in garbage out, but now we can use AI to take the trash out,” agrees Lloyds Banking Group's Rowe. “We’re already applying these solutions in critical areas, and we’ll scale them across the enterprise.”

We used to say garbage in garbage out, but now we can use AI to take the trash out. We’re already applying these solutions in critical areas, and we’ll scale them across the enterprise.

Ulku Rowe, CIO, Commerical at Lloyds Banking Group

But AI tools can help resolve some data challenges. For example, by helping to curate and refine data and understand data lineage, it can ease compliance efforts.

“There are opportunities to use AI to address data quality,” says Adam Smith, EY Americas Corporate, Commercial and SME AI Banking Lead; Managing Director, Financial Services Consulting, Ernst & Young LLP. “We helped a large bank use AI to understand and interpret data used in credit underwriting and validate whether it was correct in the underlying record. It generated a significant uplift to approximately 90%, allowing employees to focus on specific issues that are most likely to be wrong.”

These tools might still be maturing, but they can help today. Banks that adopt these tools first will find they have to invest significantly less in manually refining data.

There are opportunities to use AI to address data quality. We helped a large bank use AI to understand and interpret data used in credit underwriting and validate whether it was correct in the underlying record. It generated a significant uplift to approximately 90%, allowing employees to focus on specific issues that are most likely to be wrong.

Adam Smith, EY Americas Corporate, Commercial and SME AI Banking Lead; Managing Director, Financial Services Consulting, Ernst & Young LLP

Key actions

Objectively assess the availability and quality of data and the extent to which it’s reducing the efficacy of AI systems.
Explore how AI-powered tools can help address data issues.
Evaluate whether new data science, analytics and security skills are necessary.