AI/ML- Death of the Database? Long Live the Database!

One undeniable truth is that we often lean towards what’s familiar to us. Open-source solutions might be cost-effective and supported by their enthusiastic communities, but there’s a correlation between value and investment. Enterprise relational systems, backed by substantial licensing revenues, can afford to innovate and introduce advanced features—capabilities that sometimes surpass what open-source can offer.

However, it’s essential to understand that not every application necessarily demands an enterprise-level relational database. Amidst the evolving narrative about AI/ML’s potential to replace our jobs and the rising prominence of object storage seemingly poised to replace traditional database systems, one principle remains: “We gravitate towards what we understand.”  I predict this will happen with AI/ML, too.

There’s been limited discourse on data breaches in relation to AI since it’s heavy initial adoption. It’s not a matter of if a breach will occur; in truth, it has likely already transpired. With web-accessible tools gaining popularity, sensitive information, including patients’ medical records and residential lease details, has been exposed to platforms like ChatGPT. We must acknowledge the reality that once this data is shared with a public AI, its exposure is irreversible.

AI Needs Data

As individuals and companies grow more discerning and less influenced by the aggressive marketing of AI, they will likely adopt the same data standards and policies for AI as they do for analytics and data processing. In this transition, it’s anticipated that less vital data will be stored in cost-effective object storage, while critical data will be moved to enterprise relational systems that offer robust access controls, auditing, and data governance. Systems like SQL Server and Oracle can now internally house formats like Parquet and JSON. Moreover, they are continuously introducing advanced ML and AI capabilities within their platforms.

I predict that private generative AI products, integrated with relational databases and object storage, will become standard for most enterprise companies in the coming years. The significance of critical data remains unchanged, and we are yet to delve into how AI will align with GDPR regulations, potentially opening avenues for AI auditing businesses.

DBA 3.0

While some might perceive a door shutting, it often signifies the onset of new beginnings, paving the way for fresh opportunities, roles, and advancements. Today’s Database Administrator, now frequently recognized by titles such as Database Engineer, Site Reliability Engineer, or Database Architect, is expected to harness an even broader skill set, thereby adding heightened value. The spotlight on Data Governance is poised to shift from databases to AI, likely resulting in a pivot back to relational systems in a bid to safeguard critical data, addressing vulnerabilities left exposed by AI’s developmental phase. Historically, entrusting development as the sole guardian was perhaps an oversight, emphasizing the significance of an inclusive approach in solution formulation. This is not the first time the tech sector has navigated such waters. Numerous technologies, including what I deem as oversights in blockchain, originate from excluding essential stakeholders during the architectural stage—though that’s a conversation for another occasion.