Data Management Struggles Reveal Companies’ Challenges with AI Integration
Organizations are struggling to manage AI data effectively, with 80% reporting misuse due to employee errors, leading to data leaks. Infrastructure improvements are a priority, along with increasing storage capacity and managing unstructured data. Companies are beginning to implement data classification and governance strategies to mitigate risks associated with AI usage.
In a recent study by Komprise, data managers are grappling with a pressing issue. Four out of five organizations reported issues stemming from employee misuse of Generative AI, with sensitive data leaks becoming alarmingly common, trailing closely behind the inaccuracies often caused by AI. As companies hurry to catch up with their data management, they’re facing significant hurdles in both storage capabilities and overall administration, as noted in the AI, Data and Enterprise Risk study.
When it comes to backing up their AI initiatives, over two-thirds of the organizations surveyed view infrastructure enhancements as a critical goal. Interestingly, nine percent flagged it as their top priority after cybersecurity. More than a third—37 percent—identified boosting storage capacity as their main investment focus for AI. Moreover, just under a third—about 32 percent—stated that acquiring performant storage specifically designed for GPU compatibility was integral to their plans.
Finding and moving the right unstructured data poses a significant challenge for 55 percent of these companies. Key concerns include a lack of visibility into their existing data and the difficulties that arise from poor classification and segmentation methods. Furthermore, one in three respondents noted internal discord regarding the overall approach to data management and governance as it pertains to AI.
Krishna Subramanian, co-founder of Komprise, observed that many organizations are taking steps towards stringent AI governance and compliance. He warned that without these measures, sensitive company data risks leakage into the public domain, potentially becoming part of public large language models (LLMs).
On the tactical front, one suggested approach involves classifying sensitive data and leveraging workflow automation to mitigate misuse with AI. This tactic saw support from 73 percent of the respondents. Notably, over half—55 percent—are implementing policies and training aimed towards responsible workforce practices. While this might seem like common sense, it’s reassuring to see this shift taking place, according to Subramanian.
Some organizations are opting to limit the use of public Generative AI tools in favor of developing their internal systems. Customers are seeking enhanced visibility into their data so they can manage it more effectively. Implementing tagging, classification, and data segmentation automation will also help funnel the proper datasets into AI solutions, along with monitoring the results, Subramanian emphasized.
Interestingly, not many companies are prepared to train their own AI models at any large scale. This suggests a reduced demand for GPUs and accessible storage but indicates a need to comprehend unstructured data better. Thus, companies should concentrate on feeding pre-trained models with high-quality corporate data to drive superior business outcomes.
Subramanian pointed out that as the inferencing market matures, the emphasis will increasingly shift toward helping enterprises leverage AI effectively with their proprietary data. After all, existing models have already been built using all publicly available datasets. This highlights the importance of managing and curating data as a primary investment in AI initiatives moving forward.
The study by Komprise underscores the challenges organizations face in managing AI data effectively. With the misuse of Generative AI leading to notable concerns over data leaks and incorrect results, companies must prioritize robust infrastructure and governance strategies. Improving data visibility and classification will be essential as businesses aim to harness AI’s full potential without compromising sensitive information. As demand for effective AI usage rises, focusing on the right corporate data becomes paramount for optimal outcomes.
Original Source: blocksandfiles.com
