Christian Stegh, CTO and VP of Strategy at eGroup | Enabling Technologies.
getty
The path to AI success is paved with data quality but laden with data security land mines. In past articles, I introduced an AI maturity model. The model describes how organizations seek to advance to one of five key stages: learning, experimenting, standardizing, innovating and leading. Technology is the "how" and business outcomes are the "why" organizations are pursuing AI. This article covers a new factor: data management. This aspect is truly a journey, but it’s a worthwhile one.
Stage 1: Learning
The least mature organizations are simply getting a clear understanding of the array of data that exists and its potential value. Data sprawls across various systems, formats and locations. Unstructured data is commonly stored in multiple cloud storage services and on-premises file shares. Structured data resides in disparate databases, cloud platforms and SaaS applications. The first step is to understand the age, permissions and confidential nature of the content. This is, in fact, the current state for most of the small and medium organizations with whom we work.
Stage 2: Experimenting
As an organization inventories and understands the value and confidentiality of its data, a focus on security and privacy emerges. Robust protection measures, including encryption, access controls and compliance with regulations like GDPR and CCPA, are implemented. Typically, this entails difficult decisions to be made by business departments. It’s not IT’s role to say, "Keep this. Delete that." Rather, it’s the line of business that must decide what files to keep and protect. Data classification and retention policies are established to manage data privacy and minimize data bloat. Although essential for safeguarding data, this stage primarily focuses on protection rather than optimization for AI.
Stage 3: Standardizing
With growing data awareness, organizations begin to address data governance and life cycle management. Data stewards are assigned to own critical data assets. Another step of this stage is when an organization decides to narrow the focus of the data involved. Centralizing a small, focused dataset provides more potent ROI than general-purpose AI scanning a large unstructured dataset. Data lakes are often the mechanism to implement this centralized data store. Meanwhile, clear policies and procedures for data collection, storage, use and disposal are established, laying the groundwork for more effective AI initiatives.
Stage 4: Innovating
At this stage, organizations can focus on data quality and accuracy to enhance AI performance. Data validation processes are implemented to ensure data consistency and reliability. There are emerging technologies to assess the accuracy of large language models, but other AI models (statistical analysis, expert systems and predictive financial models) are more mature and benefit significantly from improved data quality.
Stage 5: Leading
Mature organizations foster a data-driven culture where employees understand the importance of data and its role in AI. Data literacy programs are implemented to empower the workforce. Metadata management becomes a cornerstone, providing essential context for AI model development. Continuous monitoring and improvement of data management practices ensure data remains aligned with evolving AI needs. This stage is characterized by a proactive approach to data management, driving innovation and business value through AI.
Summary
As organizations progress through these stages, their ability to extract value from AI increases exponentially. A mature data management posture provides the foundation for building robust AI models, making accurate predictions and driving informed decision-making.
Organizations that present AI tools with quality data will succeed more reliably and quickly than those trying to decipher the "garbage in" from the "garbage out." This takes time, perseverance and collaboration with data stewards in the business.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
{Categories} _Category: Implications{/Categories}
{URL}https://www.forbes.com/councils/forbestechcouncil/2024/10/25/data-managements-impact-on-ai/{/URL}
{Author}Chris Stegh, Forbes Councils Member{/Author}
{Image}https://imageio.forbes.com/specials-images/imageserve/66b3ab34e5440c366fb6121d/0x0.jpg?format=jpg&height=600&width=1200&fit=bounds{/Image}
{Keywords}Innovation,/innovation,Innovation,/innovation,technology,standard{/Keywords}
{Source}Implications{/Source}
{Thumb}{/Thumb}