Data Governance: a great ally to put limits to Artificial Intelligence

October 31, 2023

Artificial Intelligence (AI) has become a powerful tool where, based on a query about a need expressed in natural language, it builds us an answer as close and accurate as possible.

Based on this fact and understanding that the information will have been previously governed, we can ask the AI, our personal assistant and digital brain to build us a complex response to any need.

✅ For example, "build me a list of data sets that I can expose to the outside of the company based on the nature of their metadata".

Alluding to Heraclitus:

Day by day, what you choose, what you think, and what you do, is who you become.

That's what AI is all about. Day by day, learning from experience and order to gradually enter our lives and become our best and most accurate ally.

How AI can be governed?

Just like governing data, AI governance is essential to ensure data ethics and securitization. In this regard, there are several key aspects to consider:

AI should be treated as an "intelligent steward" that is given natural boundaries to prevent it from behaving inappropriately.

In fact, IBM security researchers recently managed to "hypnotize" AI models to leak sensitive financial information, generate malicious code, encourage users to pay ransoms and even advise drivers to run red lights.

In addition, just as governing use cases with advanced analytics would involve inventing processes, data sources, documenting lineage and assigning responsibilities for data, etc.

In the case of governing AI, it is, in these respects, very similar. Effective governance of AI is essential to make the most of a technology based on question-answer exchange between the user and a machine, in order to ensure that the data complies with shared social responsibility and ethical use of data in the business world.

The following figure lists some aspects that are essential for governing AI and thus avoiding these behaviors mentioned above:

How the Governance Discipline serves as a lever for AI

Having well-governed data with a foundation of structure, quality and information security will serve as an enabling lever for AI, which using Natural Language Processing (NPL) will allow for wizard queries such as:

In addition, thanks to Data Governance, the AI will be able to suggest documents with adherence agreements of the Data Sharing Agreements (DSA) or Smart Data Contracts (SDC) type for data sharing known metadata that reports or datasets make available for sharing.

It will be possible to know when a document by its nature must be certified or not, and to apply the most sophisticated security techniques that guarantee non-repudiation between sender and receiver, i.e., that both ensure confidentiality, integrity and availability and are who they say they are.

Another of the advantages of having governed data will be to enable AI to generate intelligent suggestions that improve the user experience. In this sense, thanks to proximity patterns or regular expressions, it will be possible for AI to automatically indicate which business terms in a Glossary and which physical fields in the Data Dictionary could be highly related, as well as to propose related documents or concepts based on a semantic search.

Application of AI in the DAMA framework

As a summary of the previous section, we observe a direct correspondence between AI and all the disciplines that we found in the DAMA framework.

In short, AI aims to mimic human cognitive function, but AI will be better or worse based on how we have nurtured or trained it from design. This is why, if we want an empowered AI, we must at least train it on how the company is at all levels, make it aware of the UNE specifications, ISO regulations, GDPR, Euro Orders oriented to data and AI, as well as the different white papers and existing manifestos, such as the "Data-Centric" manifesto.

It is therefore essential that as these types of tools advance and become more and more disruptive over time, the business strategy remains aligned and there is a strong sponsorship that allows to obtain the fairest expectations, providing Data Governance and AI with the best symbiosis in order to offer the best advice in their professional services.

AUTHORS
BLANCA MONFORTE RODRíGUEZ
Specialist of Data Governance Services
RAÚL HERNÁIZ ORTEGA
Senior Specialist of Data Governance Services

* * *