The topic on everyone’s lips was AI or AIOps at Mobile World Congress ‘24 in Barcelona. Indeed, it was as if that was the only topic to discuss at MWC. What happened to all the other challenges and evolutions network operators and vendors face in telecommunications? Was the pivot away from 5G a sign of frustration or an attempt at distraction from the slowed pace of 5G deployment and the need for more compelling business case(s)?
While everyone was singing a variation of the same song about how they were using AI in their initiatives and products, there wasn’t a lot of specificity to the AI initiatives and offerings being presented and discussed. Workforce automation was a frequent theme amongst vendors, and there is good reason for the well-documented labor shortage (the Telecom industry must address its talent shortage — here’s how – SDX Central) for 5G technicians and security personnel.
But AI is not a new development! In fact, it has been in development since the mid-1980s according to an ACG analyst I met with during MWC who worked on early projects for the Bell system. Those early implementations were more properly labeled as machine learning. The ML label may indeed be more accurate to describe many of the AI offerings available today. There is no question that there are many exciting large language model (LLM) projects ongoing and in the works. However, it may be the small language model projects that have the most business impact in the near term.
Solving specific business problems that utilize the raw computing power available in the cloud today guided by analytic chains may not be as glitzy as network automation with digital twins, but it has the potential to drive more immediate value and returns. Some of the more modest but useful aims of AI/ML are utilizing analytics to gain actionable intelligence for Network Personalization, Heavy User Detection, Problematic Users, Venues (Stadiums, Convention Centers, etc.), and White Glove Accounts. Using small language models for such targeted AI/ML projects can be employed more quickly and cost-effectively and yield valuable results.
It’s clear that the telecom industry is in the hype phase (or Kool-Aid stage if you prefer) for AI, and it is definitely the shiny new toy that is on everyone’s birthday and Christmas lists. Certainly, the viral success of ChatGPT (and the stock market success of NVIDIA) has been the technology trigger for this hype cycle. But will AI avoid the following trough of disillusionment from this peak of inflated expectations? Perhaps focusing on solving smaller business problems that leverage computing and save countless human hours of toil and frustration will be the stepping stone to a more successful rollout of AI.
Learn more about NETSCOUT’s service assurance Automation with AI.