AI That’s Built for Success Begins With Data That’s Ready for AI

As the use of AI in business functions becomes ever more ubiquitous, the importance of high-quality data that supports the building blocks of AI and machine learning becomes even more critical. Artificial intelligence and machine learning are dramatically changing how firms function.

To date, many financial services firms have been implementing AI in various use cases, but are primarily focused on automating basic tasks:

• User experience: AI-powered chatbots and virtual assistants provide support for users of internal technology tools, as well as client log-in
• Reconciliations: AI can significantly streamline reconciliations by using pattern recognition to perform root cause analysis
• Data outliers: AI can easily detect outliers among a set of data, such as unique swings in prices or a significant difference in transaction volume

Significant improvements in implementation would allow AI applications to be used in a more forward-looking way:

• Investment strategies: Analysis of market trends and historical data to help optimize investment portfolios
• Risk assessment and management: Machine learning models to process complex datasets to evaluate risk
• Compliance and regulatory reporting: Automation of compliance processes to reduce risk of human error and ensure adherence to regulatory requirements

Challenges in developing and successfully implementing AI

The foundation of all AI systems is only as strong as the data on which they are built. Poor quality data – data that is incomplete, inaccurate, outdated, or irrelevant – poses significant risks to the reliability and effectiveness of AI applications.

Clean data enhances the reliability of a firm’s analytics and business intelligence. With the increasing volume of data generated by firms, maintaining data quality has become ever more challenging yet essential to a firm’s efficiency and growth.

The consequences of using poor-quality data are far-reaching, including erosion of customer trust, regulatory noncompliance, and financial and reputational damage.

In addition, poor data quality can significantly affect the performance and reliability of AI systems, leading to significant issues and potential risks:

• Biased and inaccurate results: AI systems trained on poor quality data can produce biased or incorrect outcomes.
• Incorrect decisions and security risk: Erroneous or inaccurate data can cause AI systems to make poor decisions, which can have a cascading effect. In addition, poor data quality can create security vulnerabilities that malicious actors may exploit.

According to Gartner, 30% of generative AI projects are expected to be abandoned by 2025 due to poor data quality, inadequate risk controls, escalating costs, or unclear business value.

Arun Chandrasekaran, Distinguished VP Analyst at Gartner states: “Through 2025, at least 30% of GenAI projects will be abandoned after proof of concept due to poor data quality, inadequate risk controls, escalating costs or unclear business value.” ¹

Firms remain cautious about large-scale adoption of AI

Focusing on data quality in the financial services sector is crucial for ensuring compliance, managing risk, and making informed decisions. This emphasis improves operations by providing real-time accuracy and utilizing advanced tools. Although many firms are expressing significant interest in expanding their use of AI, according to new data from tech.co’s Impact of Technology on the Workplace report, caution abounds. Over two-thirds (67%) of the more than 1,000 business leaders surveyed said AI integration either remains limited or is non-existent.²

Financial services firms, in particular, are still cautious about AI’s possibilities and risk.³ Many firms are more likely to be watching and learning about AI tools rather than implementing them⁴

The exception is the very large banks, where the AI landscape is dominated by JPMorgan Chase, Capital One, and Royal Bank of Canada. For these market leaders, the path is already set, internal doubts about the quality of their data have been mostly satisfied, and a clear strategic vision has been set. That said, aside from the pacemakers, the rest of the industry is lagging, primarily owing to risk aversion.

Given the breakneck pace of adoption, it’s critical at this stage to help institutions harness the power of high-quality data and share best practices so that firms can remain competitive.

“Despite AI’s potential, most finance functions’ AI implementations have remained limited,” said Marco Steecker, Senior Principal in the Gartner Finance Practice. “As they begin to chart out a plan for how best to prioritize that additional investment, CFOs should partner with their finance leadership teams to compare their current progress against their peers’ and identify concrete recommendations from early adopters on how best to accelerate AI use in their function.”⁵

Mitigating risks with robust data management

Trusted, governed data is essential for ensuring the accuracy, relevance, and precision of AI. To unlock the full value of data for AI, firms must be able to navigate their complex IT landscapes to break down data silos, unify their data, and prepare and deliver trusted, governed data for their AI models and applications.

Continuous data quality monitoring empowers financial services companies with improved visibility across their entire data ecosystems, crucial for both operational efficiency and analytical insights. This high-quality data serves as the cornerstone for developing AI applications and training sophisticated machine learning models. By implementing a user-friendly, self-service approach, organizations can decentralize data quality management, enabling all stakeholders to proactively identify and resolve data quality issues.

The strength of AI depends on the quality of data

As organizations continue to leverage AI for competitive advantages, the focus must increasingly shift toward implementing and maintaining high-quality data management practices. By doing so, companies can reduce the risks associated with poor data, paving the way for AI solutions that are both innovative and reliable. To ensure that AI systems are reliable and responsible, data should be:

• Accurate: reflect the real world
• Complete: have required information available
• Consistent: synchronized across the organization
• Timely: available when needed
• Valid: follows business rules, in specified format, and can be used with other sources
• Unique: no instances of duplicates within the dataset

GenAI initiatives will be driven by trusted data

The key driver for any GenAI initiative is high-quality data. Since the end results will reflect the data that is being used to make predictions, that data needs to be clean, reliable, accessible, and discoverable. A well-designed purpose-built tool that integrates data quality, governance, and lineage into its design can bring a competitive advantage by giving firms the confidence that they have the appropriate inputs for large language models (LLMs) to generate responses, in addition to the right data architecture to build applications on top of GenAI capabilities.

Following are some examples of use-cases that would be appropriate for financial services and investment management:

• A GenAI-based enterprise application that can detect price movement anomalies for an investment portfolio requires a clean, consolidated dataset of instrument pricing.
• A chatbot / copilot application that can answer user questions about a firm’s investment portfolios requires curated data store with high-quality data on transactions, holdings, and so that any queries generated by LLMs can be executed and ultimately produce high-fidelity results.
• AI can simplify the extraction of insights and datapoints from unstructured data, text, and documents. Achieving this requires data ingestion and a data transformation framework that can provide AI capabilities access to the source material and then store the results.
• Using GenAI to simplify investment reporting and to handle investor inquiries requires a complete data catalog of available data fields, their semantic meanings, and sources from which data was populated.

Initiatives like GenAI represent a critical step forward in harnessing the power of AI. By collaborating with a trusted data partner, financial services firms can be confident that data used by AI technologies will uphold principles of transparency, accountability, and privacy.

Financial services firms thrive based on their reputations

Data quality plays a pivotal role in crafting effective risk management strategies and maintaining regulatory compliance. The fast-paced nature of the financial services industry means that data inaccuracies can rapidly spread across operational processes, underscoring the need for vigilance. High-quality, reliable data is essential for accurate reporting, insightful analytics, and precise forecasting in finance. Moreover, consistent data across systems and departments facilitates more informed and effective decision-making, ultimately enhancing overall business performance and reliability.

Sources

[1] Highlights from Gartner Data and Analytics Summit, March 2024. https://www.gartner.com/en/articles/highlights-from-gartner-data-analytics-summit-2024

[2] The Impact of Technology on the Workplace https://images.tech.co/wp-content/uploads/techco-impact-of-tech-workplace-report-2024.pdf

[3] 61% of Finance Execs Not Using Ai Yet: Weekly Stat. CFO.com, Nov. 8, 2023. https://www.cfo.com/news/artificial-intelligence-implementation-adoption-gartner-survey/699082/

[4] Gartner Says Most Finance Organizations Lag Other Functions in Ai Implementation, November 7, 2023. https://www.gartner.com/en/newsroom/press-releases/2023-11-07-gartner-says-most-finance-organizations-lag-other-functions-in-ai-implementation

[5] Gartner Says Most Finance Organizations Lag Other Functions in Ai Implementation, November 7, 2023. https://www.gartner.com/en/newsroom/press-releases/2023-11-07-gartner-says-most-finance-organizations-lag-other-functions-in-ai-implementation