LG Uplus announced on the 26th that it is accelerating the introduction of artificial intelligence (AI) technology to improve the quality of Internet Protocol (IP) television. The strategy aims to elevate service quality to a new level by using AI that detects and resolves issues before customers do.
On the same day, LG Uplus held a press conference at the LG Seoul Station Building in Jung-gu, Seoul, to introduce its self-developed "AI-based customer inconvenience prediction and proactive response system." This system analyzes data generated from IPTV and routers to identify signs of abnormality and proactively responds when issues arise.
Kang Bong-soo, head of the Quality Innovation Center at LG Uplus, said, "We developed and implemented an AI-based system to provide the highest quality service that customers can universally acknowledge," adding, "We aim to extend AI's application across all service areas beyond IPTV, challenging to achieve zero customer complaints."
◇ AI resolves issues before customer complaints
The "AI-based customer inconvenience prediction and proactive response system" of LG Uplus analyzes data generated by customers while using IPTV services to assess service abnormalities. The amount of customer data analyzed by AI exceeds 1 trillion data points daily.
When an anomaly occurs, AI takes the initiative in resolving it in the first instance. For example, if there is a deterioration in the quality of a live broadcast, AI can identify the issue and immediately fix it through solutions such as rebooting or remote intervention, even if the customer has not filed a complaint. Actions are taken only when the customer is not using the set-top box.
The company explained that following a pilot test, the number of customer complaints decreased by about 10%. The accuracy of predicting customer complaints was recorded at approximately 30%. The time taken from cause analysis to resolution also decreased. Data analysis, which would take about 70,000 hours if conducted manually, now takes just 6 hours, and the time taken for problem resolution has been shortened from a maximum of 3 days to immediate resolution.
◇ AI deep learning technology improves detection accuracy
The "AI-based customer inconvenience prediction and proactive response system" of LG Uplus operates in three stages: data collection, AI learning, and anomaly detection and response. It enhanced the accuracy of detecting abnormalities by utilizing the latest AI deep learning technology.
In the first stage, 'data collection', a total of over 700 types of data are collected, including device quality data generated when customers watch IPTV, as well as data related to Internet routers or network consolidation. Among these, over 270 types of data necessary for assessing abnormality are filtered and processed into a format that AI can analyze before being stored in the company's data hub.
The second stage is 'AI learning.' AI enhances the accuracy of anomaly detection by comparing and analyzing past service abnormal cases and typical cases. LG Uplus has particularly optimized the latest deep learning-based AI model, known as the 'time series data processing technology (Transformer)'.
In the final stage, 'problem detection and response', the trained AI analyzes the customer's environment to determine if the service status is normal. If abnormal signs are confirmed, it checks the device status and connection environment and decides on the most suitable remote intervention method, such as reconnection or rebooting, to resolve the issue. Through this three-stage process, LG Uplus checks service abnormalities for customers daily using optimized AI.
◇ "Not excessive data collection... goal to expand service to all IPTV customers by next year"
During the press conference, questions arose regarding the reasons for additional investment in the IPTV business, which is seeing declining revenue. The attending executives explained that they focused on improving quality to retain existing customers. Center head Kang noted, "It's important to provide a comfortable environment where customers can trust and rely on the quality of LG Uplus," explaining that improving IPTV quality would help attract more customers.
In response to criticisms about excessive data collection, Kim Jin-man, in charge of customer experience quality innovation, stated, "While I mentioned there are 700 types of data, the actual operational data consists of various patterns aggregated, making it difficult to view it as excessive data collection." He added, "Only at the final intervention stage do we differentiate between devices, but we do not continuously monitor customer information."
LG Uplus plans to prioritize applying the "AI-based customer inconvenience prediction and proactive response system" for 900,000 customers using UHD4 set-top boxes, expanding the system to all IPTV customers, reaching 4 million by next year. Additionally, the company aims to enhance the deep learning AI model used in the system to improve the scope and accuracy of anomaly detection, planning to expand the system beyond IPTV to cover home network devices such as access points.