Enterprises worldwide are eagerly embracing generative artificial intelligence (AI) as they recognize its potential to enhance innovation and unlock unprecedented efficiencies. According to recent research, a staggering 93% of organizations are already utilizing generative AI in some capacity. This surge in adoption can be attributed to the accessibility of AI tools, which are now within reach of anyone with an internet connection and an intelligent device.
While the early use cases of generative AI have shown great promise, decision-makers, including C-suite executives, are becoming increasingly aware of the risks and challenges associated with its rapid adoption. A survey commissioned by Iron Mountain, a leading provider of storage and information management services, sheds light on how organizations are using generative AI and the obstacles they face in implementing this technology.
The research reveals that half of the respondents reported using AI to create content, such as marketing or design-based input. Other significant applications of generative AI include interacting with customers through chat or voice responses, increasing team collaboration, and adding value to services and products. However, leaders also identified several challenges and risks when implementing AI.
The most prominent challenges cited by the respondents were planning for IT resources to train and implement generative AI models (38%) and sourcing, protecting, and preparing data from physical and digital assets for use in generative AI model training (38%). Ensuring the accuracy and transparency of AI models (37%) and creating and enforcing generative AI policies (35%) were also identified as significant challenges.
These concerns echo the early days of public cloud adoption, where the requirement to pay for the technology hindered enthusiasts. However, with the ready availability of free generative AI tools, citizen “data scientists” are propagating shadow AI without the necessary training, discipline, and organizational support. This lack of expertise in multiple disciplines can lead to the exposure of sensitive data, the introduction of bias, and hinder innovation rather than enable it.
To address these challenges, the research suggests that implementing a unified asset strategy is critical to the success of generative AI. A unified asset strategy enables organizations to manage, protect, and optimize digital and physical assets used in generative AI applications. It helps fill gaps and solve challenges in strategy, ethics and risk management, and practice.
Strategically, a unified asset strategy harmonizes AI initiatives and asset management while ensuring secure and environmentally sustainable retirement of digital and physical assets. It also maximizes return on investment by managing digital and physical assets involved in AI, enhancing data quality, streamlining operations, mitigating risks, and enabling flexible scalability.
Ethically and in terms of risk management, a unified asset strategy aligns policies with the organization’s goals and the nature of its assets. This alignment enables more effective policy creation and enforcement, addressing ethical use, data privacy, and security concerns.
Practically, a unified asset strategy facilitates efficient IT resource planning, allocation, and management, enabling IT teams to prepare for training and deploying generative AI models. It encompasses comprehensive lifecycle management of physical and digital assets, including digitization, metadata enrichment, and protection against unauthorized access. Additionally, it helps organizations protect and manage data and other assets created by generative AI.
While data and IT leaders agree on the importance of a unified asset strategy, the research also highlights the need for focused AI leadership. Although only 32% of organizations have onboarded someone in the role of a chief AI officer (CAIO), 94% expect this role to be filled in the future. The top priority for AI leadership is ensuring the implementation of a unified asset strategy, followed by orchestrating resource needs, following ethical practices, managing data input and output, and addressing ownership risk.