AI’s rising popularity has stirred a lot of uncertainty amongst enterprise IT teams.
Companies are set to significantly increase their investments in generative AI systems, resources, and softwarein the coming years, with their spending predicted to surge from $16 billion in 2023 to $143 billion in 2027, says IDC, a research company. However, IT crews responsible for deploying AI in organisations are apprehensive. The stakes of creating, rolling out, and leveraging AI technology for networks, infrastructure, and software development can be high, according to industry professionals.
For instance, a report published by Juniper Networks revealed that 87% of the 1,000 international executives polled feel pushed to incorporate AI tech, whilst 74% believe that their business policies are not keeping up with potential AI-associated risks and rewards. Additionally, 82% of these executives indicated that they are pressured to speed up the integration of AI into various applications.
Sharon Mandell, Senior Vice President and CIO of Juniper’s global IT crew, wrote in a blog discussing the study that, considering the rapid evolution and capabilities of AI solutions, it is understandable why the pressure to swiftly adopt AI is causing friction in many organisations. It is also understandable why policy for such potent tech often becomes a bone of contention.
While the urgency is palpable, it’s crucial to proceed with caution to avoid being left behind, says Mandell. Importantly, it’s not always necessary to completely overhaul company policies when integrating AI. For instance, most organizations already have explicit policies regarding the type of data that employees can or cannot share with third-party entities. Quite often, these policies can simply be clarified to indicate that these rules also apply to external generative AI tools.
Also, remember to consider software purchasing policies and include additional reviews for software with added AI features, says Mandell.
A study from Juniper reveals that insufficient AI networking infrastructure has led to problems with data, increased expenses, and sluggish implementation.
Juniper’s competitor, Cisco, reported similar findings in their own recent AI study. They discovered that most current enterprise networks are not equipped to cope with AI workloads. Most businesses are aware that AI will increase infrastructure workload, however, only 17% of them have networks that can sufficiently handle this increased complexity, according to Cisco.
According to Cisco, “23% of businesses lack or have limited scalability to face upcoming AI challenges within their present IT frameworks. To meet the increasing power and computing demands of AI, over 75% of businesses will need more data center GPUs to handle existing and future AI loads. Furthermore, 30% believe their network’s latency and throughput are either not optimum or below par, and 48% agree they need further progress in this area to meet future needs.”
In the words of Siân Morgan, the research director at Dell Oro Group, “businesses are aware of the necessity to use this technology for the advancement of their operations. However, amidst the seemingly limitless potential, IT leaders may be at a loss about the next concrete steps to take”. Morgan discussed this in a blog published this week, “Enterprises Brace For AI“.
Morgan suggests that businesses are just starting to create strategic plans that encompass the advantages of AI applications. “Nonetheless, investments in AIOps can be done now, which will significantly boost an organization’s efficacy,” says Morgan.
Morgan further elaborates, “AIOps utilize sophisticated analytics and ML algorithms to aid the complex tasks of network and data center operations, enhancing data center storage efficiency, forecasting network performance issues, or even automatically proposing and implementing fixes to problems.”
“The basis of AIOps is dependable input data. Network mapping ensures all IT assets are recognized, comprehended, and visualized. It also captures the relationships amongst them even when configurations alter,” Morgan penned. “AI/ML algorithms, when applied to the mix of network mapping data and real-time usage metrics, can automate various operations tasks. It might even guide the industry towards the ultimate goal of network management: closed-loop, or fully automated, operations.”
Another challenge is that AI appears very distinct from other emerging technologies of recent decades, like the cloud, Internet of Things (IoT), and mobile, according to Mandell.
“AI isn’t just about integrating a new tool or application for increased efficiency. It’s also about examining the potential impact on the entire organization,” Mandell communicated. “The fear of the unknown and the uncertainty of the outcomes make AI adoption a considerably more intricate and thought-provoking challenge for CIOs compared to other technology breakthroughs in the past.”
As per the Juniper research, some of the AI obstacles that IT teams encounter include:
