What are CSPs’ priority targets for AI deployment?

Automation is already transforming CSPs operations – and AI promises to elevate system operations and efficiencies. What approach are CSPs taking to add AI to the mix — and what are their priority targets?

Multiple trends are coming together to create fertile ground for AI adoption and deployment in the communications sector. Dynamically orchestrated 5G networks with network slicing, 6G on the horizon, edge computing and distributed cloud, modernisation of BSS/OSS, and multi-vendor ecosystem integration will all require automation and AI to ensure optimised management of the entire ecosystem.

Automation and AI will augment each other and become cross domain

Human management is not an option in such a dynamic, complex environment. So, automation is already being deployed to perform manual, repetitive tasks that are often prone to human error and require a large human resource, while bringing proven benefits and cost reductions.

Over time, different automations will be integrated together to provide cross-domain capabilities. Importantly, AI will augment that even further. AI promises to revolutionise the operations and maintenance burden on telcos, optimise the search for innovative new services, and reduce costs for CSPs, among many other benefits it promises.

As discussed in our previous blog, however, AI will not be an overnight win. Rather, it is likely to be an incremental approach. Most timelines suggest a 2026 (at the earliest) timeline for single-domain AI deployments, with 2030 widely regarded as the date AI will reach the mainstream, according to a TM Forum survey of 141 global telco industry respondents.

However, some deployments are already underway, so what are the ‘low-hanging fruits’ for early AI adoption?

Which network domains are processes are CSPs targeting first for AI?

Before moving on from the TM Forum report, it’s insightful to see which domains and which processes CSPs consider to be the most important to address first.

Unsurprisingly, Mobile RAN was the highest priority for automation and AI implementation, cited by 62% of respondents. The RAN has high operational and maintenance costs associated with it, and the scale of network elements (number of radios, cell sites, and so on). Fault monitoring and resolution, as well as energy savings, are also associated with the RAN. And, it’s just plain complicated: getting coverage right all the time is difficult and conditions change dynamically and through the seasons, impacting connectivity.

Mobile Core was a high priority for 60% of respondents overall, which is likely to be driven by support for 5G Standalone capabilities such as network slicing, as well as assuring performance for a growing number of services, with often differing demands simultaneously.

IP backbone (56%) and IP access (53%) are also seen as high priorities for automation and AI domains.

Meanwhile, in terms of processes, Fault Management was indicated across the board as a primary priority, given its pertinence to maintaining the RAN and customer experience. Fault handling is a costly and time-consuming operation, particularly if human intervention is required.

Arthur D Little (ADL), meanwhile, estimates that AI will boost customer engagement by over 15%, while reducing operational costs by up to 30%. ADL surveyed 70 chief experience officers from CSPs around the world and found that 71% emphasised improved customer experience as the primary AI benefit, including the customer-facing side of personalised sales, while 63% said they use AI to enable intelligent network optimisation and predictive maintenance. These benefits come primarily from GenAI.

GenAI is already bringing benefits to customer experience

While Gen AI holds immense promise and currently represents the leading edge of AI adoption, this is still generally confined to single domains, such as customer service chatbots. ‘Core’ AI will take longer but will bring self-healing networks, dynamic orchestration in real time, and automatic decision making. Few, if any, CSPs or telcos are at this latter stage.

In the survey, ADL identified a major telco that is already using conversational Al to reduce the number of agents and increase service quality. The chatbot helped the company to convert sales by 4x, reduce the cost per interaction by 50%, and increase customer satisfaction by 20%.

Another telco operator developed an ‘Al factory’ with a 360-degree data platform of automated marketing to offer the right product at the right time to the right customer, as well as personalised offers by using real-time behavioural and transactional data. The operator has increased sales by 6% since its implementation.

Like automation, AI will require an incremental approach

It’s clear that CSPs are embracing AI, but it will not be the ‘big bang’ adoption that the hype suggests, rather an incremental approach, beginning with single domains — and, specifically, processes within those domains.

It will be based on where the fastest value can be obtained with the least risk and disruption, maturity of the technology, growing expertise in cross-domain integration, and experimentation and lighthouse projects.

Different processes and domains will evolve at different rates, while cross-domain AI capabilities are unlikely for a few years. But the journey has started. And, We Are CORTEX can help you navigate that journey.

AI, like automation, is a moving target and requires a thorough knowledge and insight into your existing automations and domain-specific AI capabilities to ensure that the end-goal is integrated cross-domain automation and AI throughout the network and operational processes.

To read more about this journey, download our latest paper here.