More Than One-Third of Enterprise Data Is 'Dark Data,' Posing Significant Risk to Enterprise AI Readiness

Everpure and Omdia have released a joint report, "Exploring the enterprise data readiness gap." Everpure said that as the large-scale deployment of enterprise AI models and analytics expands, AI performance depends on the underlying data, and that the quality of the information companies hold is becoming increasingly important.

The report sheds light on the widening enterprise data readiness gap and also addresses the risks of "dark data," which is collected and stored but rarely reused. In this process, "dark data" emerged as a stumbling block. The title said that more than one-third of enterprise data is "dark data" and that it poses a significant risk to enterprise AI readiness.

The report said dark data includes a large amount of duplicate, outdated, and unnecessary data. It also said ROT stands for redundant, obsolete, and trivial, and noted that when low-value ROT data and high-value information are mixed together, they can increase costs and risks and may also lead to poorer business outcomes.

Companies are focusing on securing visibility, governance, and reliability to use data for AI, and are rapidly making data management a core business priority. However, dark data is acting as an obstacle to that trend.

The report said 99% of organizations have dark data. It also explained that more than half of organizations had dark data in the 51% to 75% range, and that 37% of all enterprise data was found to account for more than one-third of enterprise data.

The report warned that failing to distinguish and use high-value data can lead to wasted costs. It added that the risk of weak regulatory compliance and security vulnerabilities increases, and that opportunities to generate insights and innovation can be lost.

It also noted that up-to-date data is needed for AI systems to operate effectively. It explained that enterprises' lack of data readiness poses serious risks when applying AI to real-world operations, and that those risks are not hypothetical but are already occurring in production environments.

Organizations are facing major difficulties and obstacles in moving AI from the testing stage to the production stage. The survey found that 97% of organizations were struggling to transition from AI pilots to production environments, while 62% were facing moderate to severe obstacles.

Data management was identified as the core challenge in this process. 68% of IT leaders said data management was the biggest challenge in moving AI into production.

Storage silos and the fragmentation and growth of data were also presented as major barriers. 63% identified storage silos and the indiscriminate fragmentation and growth of data as key barriers to AI success.

The report emphasized that as AI adoption accelerates, greater visibility and control across data and underlying infrastructure are needed. It also said that success depends on knowing where the data is, how it is being used, and whether the most accurate and up-to-date copies are available.

The report pointed out that recognizing the risks of dark data and securing the visibility and control needed to manage it are separate issues.

In the survey, 76% of IT leaders recognized dark data as a major business risk, and 75% of IT leaders said AI success depends on producing actionable insights. However, 58% of organizations were found not to have basic visibility into their data environment.

The report therefore presented two data assessment criteria: business value and risk. It also advised against viewing dark data solely as a burden and said proactive mapping of data is necessary. In addition, it explained that decisions need to be made for each data set on whether to activate it, apply governance, retain it, or delete it, emphasized that this work is ongoing in nature, and said regular adjustments are needed.

Everpure proposed a framework aimed at helping organizations assess and improve their data status. The framework proposed by Everpure has four categories: High Value, High Risk; High Value, Low Risk; Low Value, High Risk; and Low Value, Low Risk.

Jung In-ho, head of Everpure Korea, said the pace of AI adoption among domestic companies is rapid. However, he said expanding AI at speed without securing data visibility is risky. He explained that companies must simultaneously respond to a complex regulatory environment, rising security threats, and cost pressures. He then described the condition for companies to accelerate innovation as an organization that can clearly answer where any data set is, who uses it, and whether it can be trusted. He also defined Data Primacy as an approach that places those three questions at the center of business strategy, and said Data Primacy points the way forward for companies.

Simon Robinson, principal analyst at Omdia, said the enterprise data readiness gap persists in most organizations. He cited a lack of visibility needed to identify and close the gap as the cause of this data readiness gap, and explained that as a result, the data readiness gap remains a risk that is not being properly managed. He added that organizations that close this blind spot have an opportunity for innovation. Vast amounts of data that had previously gone unrecognized or were fragmented can be turned into trustworthy, contextual intelligence, which can accelerate AI deployment, support better decision-making, and create greater business value.

Source: TECHWORLD · Lee Gwang-jae
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407081

References

• https://www.epnc.co.kr/news/articleView.html?idxno=407081