TL:DR;
1) There is still a lot to be gained from furthering the adoption of established data-enabled technologies such as business intelligence, predictive analytics and machine learning.
2) Both Generative AI and established data-enabled technologies benefit from a strong foundation of structured data, data governance and data-savvy employees. Even though Large Language Models enable Generative AI to make sense of unstructured data, LLMs still benefit from structured data.
3) The consumption of LLMs through tokens can be expensive. There is a risk of organisations over-using LLMs for routine queries that could easily be served more efficiently and cheaply by search engines.
4) Organisations have done a lot of work to establish good governance over their existing data-enabled technologies, but governance of Generative AI is still in its formative stages, and the risks of data leakage, data privacy breaches and hallucination by LLMs still need a lot of work to address.
The rise of Generative AI (GenAI) has sparked conversations about the future of technology in business, leaving many wondering if traditional IT systems—particularly established data-enabled technologies—are becoming obsolete. While GenAI presents exciting opportunities, there’s still much to gain from further advancing and adopting technologies such as business intelligence, predictive analytics, and machine learning. Rather than viewing GenAI as a replacement for these systems, organizations should focus on building a symbiotic relationship between them, underpinned by a strong data foundation and robust governance practices.
1. Established Data-Enabled Technologies: Untapped Potential
Before diving headfirst into the possibilities of GenAI, it's essential to recognize that many organizations have yet to fully leverage the potential of traditional data-enabled technologies like business intelligence (BI), predictive analytics, and machine learning (ML). These technologies have long been integral to business decision-making, enabling organizations to extract insights from structured data, optimize processes, and predict future trends.
BI tools have evolved, becoming more intuitive and powerful, allowing businesses to unlock hidden patterns in their data. Predictive analytics, when combined with machine learning, can offer even more granular foresight into customer behavior, supply chain optimization, and financial forecasting. Yet, many companies are still in the early stages of harnessing these tools effectively, leaving significant value on the table.
Generative AI may be the talk of the town, but it doesn’t eliminate the need for these more established technologies. Rather, the two can coexist, with GenAI complementing the insight-driven capabilities of BI and ML. Both types of tools are essential for different use cases, and organizations should strive to maximize the benefits of both.
2. Data Governance and Structured Data: The Backbone of AI Success
One critical misconception about GenAI is that it can bypass the need for structured data. While it's true that Large Language Models (LLMs) like ChatGPT are trained to make sense of unstructured data—such as natural language text or images—their effectiveness still improves with access to well-structured, clean, and labeled data. Structured data enables more accurate predictions, helps train better models, and enhances overall AI performance, regardless of whether it is traditional machine learning or cutting-edge GenAI.
Moreover, the success of any AI initiative, including GenAI, is built on solid data governance. Proper data management practices ensure that data is reliable, secure, and accessible. Without a foundation of strong data governance, even the most advanced AI models can deliver inaccurate results, suffer from bias, or create compliance risks. Data-savvy employees, who can navigate this ecosystem of data governance, are invaluable in both traditional and AI-powered initiatives.
While GenAI's strength lies in generating new content, it can’t replace the structured, analytical frameworks that drive predictive models and business insights. It is the synergy between structured data and unstructured data interpretation that leads to the most powerful AI applications.
3. Cost of LLM Consumption: A Case for Balance
One of the major challenges associated with GenAI is cost—specifically, the consumption of Large Language Models through tokens. Each interaction with a GenAI model, such as querying it or generating text, consumes tokens, which in turn incurs costs for the organization. This can quickly add up, particularly when LLMs are used for tasks that could be handled more efficiently by traditional technologies.
For instance, routine queries that search engines or BI tools could answer—like finding basic information or simple data lookups—may not require the capabilities of a full-fledged LLM. The risk here is that companies might overuse GenAI for low-complexity tasks, unnecessarily inflating costs. For simpler queries, search engines or traditional BI tools can provide faster, cheaper, and equally effective solutions.
Striking a balance between using GenAI for complex, creative tasks and leveraging traditional tools for more straightforward queries will be key to keeping costs manageable while maximizing value.
4. The Governance Gap: Traditional IT vs. GenAI
While organizations have made significant progress in governing their traditional IT environments, including BI, predictive analytics, and machine learning, the same cannot be said for GenAI. The governance of GenAI technologies is still in its infancy, and the risks are numerous. Data leakage, privacy breaches, and hallucination (where AI generates incorrect or nonsensical outputs) are just a few of the pressing concerns that organizations must address.
Traditional IT systems have established frameworks for governance—ensuring compliance with data privacy regulations, protecting sensitive information, and controlling access to data. But with GenAI, new challenges arise. Since LLMs rely on massive datasets, some of which may include proprietary or sensitive information, the risk of unintended data exposure is high. Additionally, the unpredictable nature of AI hallucinations can lead to false conclusions or inappropriate content generation, posing reputational risks.
To safely incorporate GenAI into their operations, organizations need to invest in new governance frameworks specifically designed for AI technologies. This includes ensuring data used by LLMs is properly vetted, setting boundaries on what LLMs can and cannot access, and continuously monitoring outputs for quality and accuracy.
Conclusion: A Hybrid Future for IT
The rise of GenAI doesn’t signal the death of traditional IT systems or established data-enabled technologies. Instead, it opens the door to a hybrid future where organizations can harness the power of both. Business intelligence, predictive analytics, and machine learning still have immense untapped potential, and a strong foundation of structured data and governance is critical to the success of any AI initiative.
By striking a balance between using GenAI for advanced, creative tasks and relying on traditional IT systems for more established, efficient processes, organizations can ensure they are making the most of their technology investments—without falling into the trap of over-relying on costly, nascent technologies like LLMs.
In this evolving landscape, the key to success will be thoughtful integration, robust governance, and an unwavering commitment to making data the bedrock of business operations.