Alex Choi, AI RAN Alliance & SoftBank (00:21):
So far, AI/ML has primarily been used in the network analytics, but going forward, we expect its focus to shift towards network management automations, especially with the reasoning capabilities demonstrated by the latest large language model and soon to be introduced AI agent features. We will likely see automation, network automation systems evolving to minimize the need for human operator intervention in network management. Moreover, this shift will have measurable effects such as opex saving that can be quantified.

Patrick Kelly, Appledore Research (01:10):
You see five big categories for operators to leverage ai. One is in the network. This is for network dimensioning and optimization. The second area would be in service assurance. This is for root cause analysis and things like traffic forecasting. Third area would be network security for things like threat detection, digital enablement, and finally internal efficiencies.

Ronnie Vasishta, NVIDIA (01:43):
So the telecom networks are actually a fertile and very rich area for AI because they have tremendous amount of data that's being collected off the network. Also, whether it be user data or whether it be network data, we see AI being used both in the signal processing, but also in the network operations. And how do you manage operations for, say, energy efficiency, for ensuring that we write the quality of experience, especially as we start deploying generative AI or AI applications at the edge. So an incredibly rich and fertile area for use of AI in the telecom networks.

Phil Cutrone, HPE (02:31):
Well, initially, I think the operators will continue this journey down cost reduction and network resiliency initially, but there's a lot of new emerging features that are already starting to come through. Some of the operators, for example, call recording or call summarization, and to some point real time translation. It's hard to get there, but we're seeing translation with messages already. And the aspiration is to get to real time translation in the future, a lot of work ahead of us. And then finally, we want to step back and think about new revenue streams for the operators. That's really the compelling opportunity here, not just keeping future competitiveness, and that is doing something around the big opportunity around enterprises. So we have large language models that are being tuned by enterprises that need low latent inferencing at the edge. And I think that presents a tremendous opportunity for telco operators to actually participate in AI workloads by offering inferencing services at the edge close to their RAN site.