Generative artificial intelligence (GenAI) is sure to unlock a plethora of perks for telecom operators, but they will need to remain focused on proximate benefits while investing in the larger infrastructural opportunities still out of reach, according to two principal analysts at Analysys Mason.
AI hype has quickly outpaced enthusiasm for the metaverse since ChatGPT’s release in November last year, analyst Martin Scott wrote in unreleased research shared with Silverlinings.
“One key reason for the high engagement is that generative AI can be ‘played with’ right now and take-up has been rapid. This means it can potentially be monetized much more rapidly than metaverse use-cases which may not gain mass take-up until 2030 or beyond,” Scott wrote.
AI’s surge in the industry was further reflected in a recent Gartner survey, which highlighted “hesitant, but not frozen” C-suites citing AI as the most significant technology to impact their industry over the next three years, according to Gartner VP and Analyst Kristin Moyer.
Yet while GenAI will present a long line of benefits to CSPs — from customer experience to software development — significant barriers remain, particularly when it comes to current data architectures and governance, recent research from Analysys Mason Principle Analyst Adaora Okeleke explained.
Training models for GenAI will take time
GenAI models demand billions of data points across various data source locations and formats to be trained. But “access to and the management of high-quality data remains the bane of AI projects within telecoms,” according to Okeleke.
The training process for these models also requires accessing text-based data sources, which are often expensive and time-consuming challenges. “More importantly, CSPs use several network equipment vendors. Managing and aligning data from these providers for GenAI will be challenging,” she wrote.
Privacy and bias concerns associated with GenAI also require improved data governance practices from operators, which becomes increasingly difficult with the uptick in data volume and distribution necessary for integrating GenAI.
“The data catalogue, a strategic asset when it comes to data governance, is now playing an important role in helping CSPs define what data assets they have [and] how these data assets are used across the organization,” she explained in an interview with Silverlinings. “This function will be important for GenAI use cases and defining who exactly should be having access to the data.”
This privacy process will take all industry stakeholders working closely together “to define and implement the required guardrails required to avoid threats such as the infringement of customers’ data privacy from occurring,” Okeleke’s research further described. That includes everyone from vendor partners to cloud providers.
Focus on the present (use cases)
While AI remains a nascent technology to telcos and other industries alike, Okeleke said industry collaboration will be key for effective early use cases.
She expects GenAI-based chatbots to be among the first use cases deployed with software developer collaboration. From there, she believes the next stage will be data collection that can then enable using GenAI to target specific business challenges.
“Next steps would then be deciding the right environment to store and manage the relevant data sets,” she explained in the interview. “I expect this to be a combination of both hybrid and multi-cloud environments as CSPs will hold some data sets on-premise or in their private cloud environments and on multiple public cloud environments,” which will also be essential in ensuring traceability in GenAI models.
This sentiment is shared by fellow principal analyst Scott, who reasoned that operators should focus more on the immediate, consumer-facing benefits now, or they risk missing the “fast-follower” train that operators typically jump on.
“Telecom operators should consider this technology as an opportunity to differentiate themselves and create new value for their customers. Generative-AI-based solutions could be released by operators’ competitors tomorrow,” Martin wrote in the unreleased research.
Solutions like improved customer chatbots “can rapidly enhance operators’ existing plans for the digitalization of customer service,” he wrote, adding that “personalized marketing and sales should also be prioritized as it can potentially improve customer engagement and satisfaction while increasing revenue.”