Many telcos nonetheless underestimate gen AI’s broader potential, warned Frost & Sullivan Senior Business Analyst Soumyadeep Roy Chowdhury
In sum – what to know:
Not magic — foundational – Gen AI can dramatically enhance telecom operations, from sooner community troubleshooting to fraud detection and predictive automation, however solely when operators have clear information, unified programs, and powerful governance in place.
Larger than chatbots — however not plug-and-play – Telcos danger each underusing and overhyping gen AI: limiting it to chatbots misses main positive aspects in documentation, design, and state of affairs simulation, but anticipating it to repair fragmented OSS/BSS or poor-quality information will result in disappointment.
In a latest dialog with RCR Wi-fi Information, Soumyadeep Roy Chowdhury, senior business analyst at Frost & Sullivan, outlined how generative AI (gen AI) is reshaping telecom operations. He highlighted the faster-than-human velocity with which the expertise can detect and resolve community points, in addition to how gen AI is changing into a catalyst for adjoining applied sciences corresponding to augmented actuality and AI-powered fraud detection — finally creating intelligence that “extends far past easy optimization.”
“This convergence alerts a future the place telecom operations evolve from reactive guide intervention to proactive, autonomous, and predictive programs pushed by AI-enabled collaboration throughout a number of expertise domains,” he mentioned.
The monetization outlook for gen AI can be extra optimistic than, say, 5G, which Chowdhury identified required heavy infrastructure funding and lengthy, unsure payback cycles. “Generative AI is totally different because it delivers speedy, measurable outcomes with far decrease entry limitations, primarily pushed by software program, information, and course of integration quite than massive capital upgrades,” he defined. Moreover, by immediately impacting revenue-focused enterprise areas corresponding to buyer expertise, advertising, and fraud prevention, gen AI can result in important operational financial savings.
Nonetheless, he warned that many telcos nonetheless underestimate generative AI’s broader potential. “Some telcos and related companies are limiting GenAI investments to chatbot functions, whereas it has higher worth propositions in duties corresponding to automating documentation, RFPs, technical designs, ticket summarization, and root-cause narratives, and information retrieval,” he mentioned. These capabilities can “considerably scale back cycle occasions” throughout engineering, operations, gross sales, and procurement.
One other main alternative lies in state of affairs simulation. In keeping with Chowdhury, gen AI may help workers mannequin service demand, community visitors patterns, churn, tariff impacts, or rollout sequences — providing “what-if views” that help extra knowledgeable decision-making.
Nonetheless, he cautioned towards overestimating the expertise: It received’t “magically flip issues round or utterly remodel” a telco’s operations, he mentioned. “We have now to grasp that gen AI, like some other AI answer, is a significant transformative expertise that acts as an enabler. It doesn’t utterly take over the possession of an entire workstream, however certainly, can remove some [manual] processes with automation.”
A key problem, he added, is that some operators assume gen AI will ship outsized advantages even in environments with fragmented OSS/BSS, poor information high quality, or siloed programs. “However realistically, with out information unification, governance, and APIs, gen AI is more likely to disappoint,” Chowdhury warned, including {that a} utterly unsupervised gen AI system can result in hallucinations. “A completely autonomous community powered by gen AI will not be one thing that might be achieved in a single day; it’s most undoubtedly not a plug-and-play expertise.”
To get there, operators should first set up high-quality information integration, governance, and APIs. Scaling gen AI will then require cloud, hybrid, or edge architectures. “Scalable compute, distributed processing, and automatic mannequin administration are wanted to deal with large real-time information hundreds, keep efficiency at scale, and constantly replace and monitor AI fashions in manufacturing,” he mentioned.
