EY: How CIOs Can Build Gen AI at Scale While Managing Risk

In this deep-dive, we speak to EY’s David Kadio-Morokro on ways CIOs can build Gen AI capabilities at scale while managing risk

Following his EY report, Financial Services CIOs – building Gen AI at scale while managing risk, Americas Financial Services Innovation Leader for EY, David Kadio-Morokro, speaks to FinTech Magazine about the challenges CIOs face when looking to scale Gen AI, and how these challenges can be overcome. 

Of course, to effectively build new Gen AI capabilities at scale, Chief Information Officers (CIOs) must first see the value of unlocking these new capabilities across the tech stack – for driving efficiencies, reducing cost and transforming banking operations. 

The value of Gen AI across the tech stack

For David, while it’s clear many CIOs see the value in Gen AI implementation, new capabilities can be hard to action

He explains: “Financial institutions continue to maintain sizeable traditional technology portfolios with an emphasis on structured data movement from Front Line Units (FLU) to middle and back office. 

“Incremental investments in high-speed and unstructured data processing are often decoupled from their existing infrastructure, leading to data proliferation and frictions in data usage.”

To provide banking associates with the right Gen AI tools, whether for engaging clients, managing risk or compliance, David recommends CIOs invest in the following priority capabilities: 

  • Unstructured data processing at scale: doubling down on investments in data discovery, tagging and classification tools to uncover latent metadata within unstructured documents.
  • New data storage mechanisms: expanding investments in graph and vector databases that store metadata (e.g., entities, relationships, properties) embedded in the documents to support systematic querying and analysis on unstructured data. 
  • LLM fine-tuning: investing in capabilities to support fine-tuning and tuning of large language models (LLMs) to expand the boundaries of their applicability for firm-specific use cases.

The challenges to building these capabilities

David Kadio-Morokro, EY

While the roadmap to deliver Gen AI change is clear, there are challenges that CIOs must first overcome to achieve the three key aims outlined above. 

Firstly, “the relative infancy of the agenda presents a major challenge in that this is a new frontier for CIOs across the board,” says David, “bringing a plethora of learning curves along with it.”

He continues: “For example, it’s difficult to manage the voracious appetite of LLMs for data. The applicability of increasingly diverse data sets for Gen AI use cases increases the risk of inadvertent proliferation and exposure, leading to potential data security and privacy issues. Conversely, a lack of availability of data for use cases can stifle innovation with lost opportunity cost.

“We’ve also seen a new phenomenon we call 'vendor confusion.' Hundreds of vendors are offering models, data stores, libraries and other assets, which makes it hard to navigate and determine the combination that works best. There is also a lack of agreement on the highest-value use cases, which results in long periods of testing.”

Gen AI: A people, process and technology triumvirate

At its core, David says to successfully implement Gen AI, CIOs and IT operations need to successfully balance people, processes and the technology itself to maximise impact. 

“A successful implementation of Gen AI requires cross-collaboration across all layers of the organisation,” David notes. “It involves finding and training individuals with AI expertise, adjusting how work is done to use AI most effectively and adding new AI tools into the current technology stack.

“CIOs must ensure that their workforce is properly trained to design, build and use Gen AI tools – such as copilots – effectively to support their daily work in a safe manner. 

“Upskilling employees whose roles may have been impacted by Gen AI ensures they can focus on higher value work and creates further opportunities within the organisation. Leadership should champion the initiative to embed Gen AI broadly within the institution.”

He concludes: “It’s equally as important for CIOs to update processes to incorporate Gen AI tooling to add value, and also to create and change processes to document the critical knowledge held by employees. 

“Additionally, CIOs must ensure they evolve risk management and compliance processes to measure and mitigate new types of risk.

“Finally, from a tech perspective, designing, building and scaling Gen AI tooling will support the changing needs of the enterprise in a safe and controlled manner.”

**************

Make sure you check out the latest edition of FinTech Magazine and also sign up to our global conference series – FinTech LIVE 2024

**************

FinTech Magazine is a BizClik brand.

Share

Featured Articles

Jason Maude, Starling Bank, Joins FinTech LIVE London

Jason Maude, Chief Technology Advocate at Starling Bank, to speak at FinTech LIVE London – 22 and 23 October 2024

FinTech LIVE New York – Reasons to Attend

Returning June 2024, FinTech LIVE New York will be the ultimate event to connect the world’s influential fintech leaders in North America

FinTech LIVE New York – One Month to Go!

Just one more month to go until FinTech LIVE New York, don’t miss out on attending the ultimate virtual event for fintech leaders in North America

The Global FinTech Awards: The Categories – Part 3

Financial Services (FinServ)

SUBMISSIONS EXTENSION: The Global FinTech Awards 2024

Tech & AI

Joel Perlman, Co-founder, OakNorth joins FinTech LIVE London

Financial Services (FinServ)