Creating Data through AI
The Intersection of Data Governance and AI Governance – DATAVERSITY
Date:
Artificial intelligence (AI) is revolutionizing enterprises and sectors across the globe through innovative data products and services. According to a recent AI Index Report by Stanford University, approximately 55% of companies have adopted AI in at least one department or function. For optimal outcomes, businesses must align AI and its data operations with their strategic goals. Therefore, data governance (DG) and AI governance (AIG) are essential.
Data governance aligns data-related tasks within companies, and 88% have already set up a program for it. AIG, being more recent to this, is concentrating on overseeing machine learning (ML) algorithms and other artificial intelligence (AI) systems to drive profits and guarantee their use is both fair and compliant with regulations.
With the growing need for both data governance and AI governance, how can businesses integrate these frameworks effectively? This article delves into their areas of focus, commonalities, distinctions, and the strategies worth considering.
Understanding Data Governance
Data governance refers to a collection of procedures, responsibilities, guidelines, norms, and measurements designed to guarantee the proper and efficient utilization of data within a company. This initiative encompasses overseeing the planning and administration of data, which the American Heritage Dictionary defines as "facts that can be analyzed or used to gain knowledge or make decisions." Effective data governance enables organizations to rely on and leverage their data confidently.
The primary goals of data governance are accomplished through teamwork and cooperative governance. It strives to meet the organization's requirements for insights derived from data, facilitate the effective application of technology, and promote digital transformation efforts. Refer to the diagram below:
As a result, those in charge of data governance make choices, including:
These choices affect the way AI is trained and how it generates content. For instance, AI might determine, create, and utilize organizational data that is regulated. These outcomes influence important data governance outputs. Nonetheless, AI projects are only a portion of data governance's responsibilities.
Data Governance Extends Beyond AI
Data governance oversees the requirements for data across all company technologies and staff involved in data handling. AI is merely a component of this broader framework, which also involves adhering to data regulations such as the EU’s GDPR. Physical systems with capabilities beyond AI encompass:
Additionally, Data Governance (DG) primarily emphasizes the importance and impact of data, avoiding technical specifics that are less pertinent to business professionals. For instance, DG dialogues might concentrate on the standards required to protect data using encryption. Nevertheless, delving into which specific encryption algorithm to implement or how to modify it with ENCRYPT-CSA generally falls outside the realm of DG.
These elements of data governance indicate that DG oversees the data requirements from all company technologies and staff who work with data. Artificial intelligence projects are only a single facet of DG's focus. As AI technologies advance, businesses also require specific AI governance structures to address the unique issues related to their creation and application.
What Does AI Governance Entail?
AI governance (AIG) manages the procedures, responsibilities, and technologies that form the foundation of a computer's cognitive abilities, which mimic human intelligence, extending beyond mere data.
These elements encompass system design, monitoring, and risk management.
Although the structures of AIG may differ, they all share the goal of fostering comprehension, responsibility, and openness in AI creation, progression, and results. Within this framework, AIG provides guidelines to help businesses harness the benefits of AI projects while guaranteeing that AI technologies and systems are secure and morally sound.
To achieve success, those leading AI governance efforts should take into account
Obviously, AIG goals will overlap with those of data governance. Both will have to guarantee that any training data used for AI and any resulting data products are in line with business requirements.
Furthermore, AI administrators must keep track of all data processed by AI systems and monitor its usage. Nonetheless, AI oversight extends beyond just managing data elements.
AI Governance Extends Beyond Just Data
An AI governance framework should not only regulate data practices but also set guidelines for system architecture. For instance, AI governance needs to oversee and evaluate both the advantages and potential risks associated with AI performance. As AI systems become more advanced, increased governance is essential to ensure they meet safety and ethical standards. The various types of AI applications include:
Although AI systems with advanced intelligence can provide substantial benefits to businesses, they also elevate the potential hazards to human safety and health. This analysis introduces a viewpoint to Artificial Intelligence Governance (AIG) that is absent in Data Governance (DG). For instance, any regulatory framework should adopt a risk-based strategy, as illustrated in the following model, to align with the EU's AI Act.
AI governance must address the nature of the data inputted into and outputted by AI systems, alongside evaluating the sophistication of the AI itself. This approach helps to tackle challenges such as bias, privacy concerns, intellectual property rights, and the potential for technology misuse.
As a result, AIG must regulate the topics that AI can handle, the timing, and the circumstances. For instance, if an AI tool creates a list of potential job applicants, the AIG guidelines should provide recommendations to guarantee the list is fair and used properly.
AIG and DG have overlapping duties when it comes to managing data as a product used and generated by AI systems, even though they are distinct in some aspects. Both governance initiatives focus on assessing data integration, quality, security, privacy, and accessibility.
As an example, both governance structures should guarantee that high-quality information aligns with business requirements. If a large retail company found that their AI-driven recommendation system was proposing unrelated products to consumers, both Data Governance (DG) and AI Governance (AIG) would aim to fix the problem.
Nonetheless, the optimal way to address the issue might involve using either method alone or a mix of both. Identifying the appropriate governance response involves examining the underlying problem.
Contrasting AIG and DG
AIG and DG offer distinct methodologies, with their effectiveness varying based on the specific issue at hand. Consider the scenario mentioned earlier, where a customer receives incorrect pricing details in response to an inquiry.
The data governance team reviews the product data pipeline and discovers that there are inconsistencies in data standards and some attributes are missing, which are affecting the AI model. Meanwhile, the AI governance team notices that there are chances to improve the recommendation algorithm by better adjusting how customer preferences are weighted.
The retailer could address the data quality problems using data governance, while AIG enhanced the functionality of the AI model by adopting a collaborative strategy that incorporates both data governance and AI governance viewpoints. This joint endeavor could eventually deliver more pertinent and valuable product suggestions to customers.
However, the organization might lack sufficient data to identify the underlying issue precisely. Leadership could be presented with either hypothesis or another explanation, such as a network error caused by a security measure. In these situations, the methods of DG and AIG might or might not effectively resolve the issue.
Summary
In a podcast episode, Karen Meppen, who serves as the Director of Client Services at Hakkoda, proposed that it's important to thoroughly understand the context of a business goal to determine if governance should be the subsequent action and to identify the type of governance required. The inquiries necessary to grasp a business goal might be straightforward and could indicate the need for either a data governance strategy, an AIG strategy, or a mix of both approaches.
Although DG and AIG frameworks share similarities, they remain separate entities with unique goals and results. Often, the most effective approach for an organization is to gather additional information about the data issue or opportunity prior to determining if it falls under the DG or AIG category.
Meppen mentions, "You can make various modifications to your dataset that might result in a series of negative effects. At the same time, certain data manipulations can produce a chain of beneficial results."
The key is to strategize for potential unknown problems and to gain insights into any data issues that arise. Grasping the problem is crucial to determine if a data governance or AI governance framework should be implemented and how to do so.
Wholesale CBDCs are expected to be introduced sooner than retail ones, according to BIS.
EBAday 2024: Examining the hurdles and advantages of CBDCs
Latest Insights
Long-term Bitcoin holding patterns are increasing the illiquid supply.
Concerns from the SEC about USDC could complicate Circle's plans for an IPO, reports Barron’s.
JP’s Top Product of the Week — Today at 4pm Eastern! 6/18/24 @adafruit #adafruit #newproductpick
Google collaborates with Harvard to create the most detailed map of the human brain to date #ArtTuesday
PlayStation Store Update Global June 18, 2024 – PlayStation LifeStyle
Echoes of Wisdom's Zelda without a sword focuses on gameplay quality over canonical elements.
Copyright @ 2024 Plato Technologies Inc.
