AI: when technology isn't everything

AI: when technology isn't everything

We are witnesses, as a natural result of the growing adoption of information technologies, that AI has become a commodity for companies.

The facilities offered by the cloud have democratized access to this technology, allowing almost any organization, even small ones, to leverage its benefits.

Thus, just as the Internet transformed the way of doing business, producing, and consuming, AI is becoming an essential element for corporate operations at any point in the value chain.

Unlike the transformation produced by the internet, where organizations had to undergo digitalization for their survival, AI has not yet exerted that pressure; instead, it has positioned itself as a differentiating element, optimizing efforts, generating new ways of understanding operational indicators, reducing risks, and anticipating our customers' needs.

The incorporation of artificial intelligence, although it has become easier, is not a switch that generates transformation when flipped; its impact is almost directly proportional to two aspects with which companies are struggling: data governance and data quality.

Data governance is a capability that companies must cultivate; it is closely related to the understanding and valuation of data management, and the way in which business and technology areas operate together in that management.

This further challenges organizations because it is an evolutionary capability; it cannot be created technologically, although it can be supported by it.

Clear definitions of responsibilities in any data governance scheme are not sufficient if there is no genuine conviction from everyone to exercise them and ensure their compliance over time.

It is essential for every member of the organization to develop the ability to handle information and data in a secure, ethical, and controlled manner. As a valuable shared resource, it is everyone's responsibility to understand and ensure its correct management and utilization.

And if data governance is an organizational capability, data quality is the realization of that governance. We could not have good data quality without committed governance capable of managing them.

Data quality in itself poses a major challenge; it is a promise for the future that manifests at the final stage of the consumption chain, when what can be done to improve it is almost nothing.

If we consider quality attributes as intrinsic to data, the quality of artificial intelligence solutions is intersected by those attributes. And, unfortunately, no AI model or solution can escape this problem when it already exists, and worse still, its findings or recommendations may be poisoned by that lack of quality.

According to the Journal of the Knowledge Economy article, "Overview of Data Quality: Examining the Dimensions, Antecedents, and Impacts of Data Quality," one of the causes affecting data quality is the lack of accountability from management areas.

Although it is emphasized that, at the same time, they are the ones who reap the benefits of good data quality—for example, by making informed decisions based on trust in the data. Another discovery that emerged from this analysis is that metadata quality has a significant impact on the perception of general quality and on the opinion of whether or not a piece of data possesses quality.

Consequently, the success of an AI solution is fundamentally based on data quality, not just as a general characteristic, but also in conjunction with its business context information, cycles, and periods of validity or completeness, among others. Furthermore, the assessment of an AI solution's impact for users is directly related to trust in the quality of the data.

Of course, not all is lost.

Nowadays, solutions already exist to resolve quality issues, but once again, they require definitions, metrics to understand their current status, the capacity to analyze the situation, and the ability to act through the implementation of continuous and measurable initiatives.

All of which is the result of applying a governance framework. Data quality would not be lasting without data governance to manage it, and this is where, as mentioned, there are no magic solutions.

The sooner this importance is recognized and a long-term commitment is made to promote an organizational data culture—one that includes incentives for improvement and better communication regarding the value and utility of data, without missing the opportunity to celebrate management successes—the sooner we will begin to experience the improvements that artificial intelligence solutions can provide to individuals and organizations.

The path toward the implementation of high-impact AI solutions must have a comprehensive, cultural, continuous, and interdisciplinary approach to ensure that organizations can unlock the potential of their data and turn it into strategic allies, fulfilling the promise of this new technological era.