The adoption of artificial intelligence (AI) in the telecommunications industry continues to face challenges due to fragmented deployment, preventing operators from fully realizing the technology’s potential in improving operations and customer service.
Despite clear benefits, such as improved network optimization and more personalized services, AI implementation often remains limited to isolated functions. According to Genie Yuan, regional vice president for Asia Pacific at Couchbase, this piecemeal approach prevents meaningful transformation.
“Traditionally, network optimization has been reactive,” said Yuan in an email interview with Back End News. “Operators wait for a surge in demand, analyze the pattern, and then optimize the network accordingly.”
Yuan explained that the goal is to shift toward a more automated process where the network adjusts itself in real time based on predicted demand. AI can help prevent congestion and outages by dynamically adjusting network parameters.
Downtime and network disruptions remain major concerns for telecom companies and customers alike. Yuan pointed out that AI can be used for real-time anomaly detection and predictive maintenance.
“Most of the time, anomaly detection is done using statistical models such as regression models that help us analyze patterns based on past incidents,” said Yuan. “AI helps process streaming telemetry data to anticipate failures. From my perspective, preventative maintenance is becoming an exciting area.”
Yuan also discussed the role of Agentic AI, which integrates data from various sources to support real-time decision-making and provide more tailored services. This technology supports predictive analytics that helps match user demand with network capacity.
“When you map this demand side to the infrastructure supply side, it gives you a much clearer picture of overall network performance and helps optimize the entire system,” said Yuan.
However, widespread transformation is held back by the way telcos currently deploy AI. The technology is often introduced in specific areas rather than being rolled out across all operations. AI depends on real-time, high-quality data to be effective, and without integrated systems, its benefits remain limited.
“Until we see more widespread AI deployment across entire operations, AI use will likely remain fragmented, focusing on specific processes, not full transformation,” said Yuan.
Responsible use of Agentic AI
Another emerging issue is the responsible use of Agentic AI, especially given its ability to reason. Yuan stressed the importance of strong data governance.
“This starts with centralizing data access control and audit logging, enabling operators to track who is accessing what data and when,” Yuan said. “It’s also critical to ensure that data remains within the legal jurisdictions of where it originated.”
As AI becomes more common, telecom companies must implement Responsible AI (RAI) frameworks that include explainability and fairness tracking. Operators will also need to manage both human and AI agent identities, especially across hybrid and multi-cloud systems.
Deploying AI at scale requires strong security foundations to protect data and ensure compliance. Yuan noted that while the use cases evolve, the core principles of security and governance must remain consistent.
Legacy systems also contribute to delays in adopting AI. Many existing infrastructures lack the capacity to support real-time or scalable AI workloads. Data silos, outdated data practices, and centralized systems limit the scalability needed for technologies like 5G and IoT.
To address these challenges, Yuan suggested using distributed platforms that support transactional, analytical, mobile, and AI workloads; adopting cloud-native architecture for flexibility and elasticity; and replacing legacy systems with real-time data synchronization and processing.
In the Philippines, natural disasters frequently damage network infrastructure, making edge computing critical. A distributed setup with edge capabilities helps maintain operations during outages and ensures continued access to services in times of crisis.
“From an AI perspective, foundational models can drive major improvements in the telecom sector,” said Yuan. He noted the multilingual customer service bots and global anomaly detection models as examples that would work well in the Philippine setting, where linguistic diversity and infrastructure challenges persist.
Get the latest before it trends. Follow Back End News on LinkedIn, Facebook, X, YouTube, and TikTok for updates and in-depth coverage across the tech and security landscape.
You must be logged in to post a comment.