The Telco AI Journey: Challenges, Suggestions & Current Progress
In part three of this three-part series on AI, we discuss examples of telecom operators that have implemented AI in their operations and services and look at the challenges they face. Read parts 1 and 2 here:
1. AI: Understanding What It Means & What the Future Holds
1. AI: Understanding What It Means & What the Future Holds
The telco industry has historically been at the vanguard of trying out new technologies and creating new services and applications that incorporate such technologies in their portfolios. The advent of AI has similarly not gone unnoticed by telcos, with many attempting a variety of new use cases to harness the potential that AI affords.
AI-assisted use cases best suited for telcos
Although there are several use cases that involve telcos using AI, a few stand out as having the maximum possible impact. Some of these are:
- Network optimization è Telcos can use AI tools to proactively study and track their networks and other data parameters to identify potential problems before they occur. They can then plan preventive steps for fixing network problems, predicting network congestion, and proactively rerouting traffic to avoid network outages.
- Personalized customer services è AI-powered customer service chatbots are becoming commonplace in telcos, with many already launched. Bots offer personalized experiences to users and can boost efficiency and reduce costs. An IBM study shows that companies can save US$1 million annually for every second they shorten the average call center handing time. Furthermore, AI-powered chatbots can help reduce average query handling costs from US$5-12 to US$1.
- Sales & marketing enhancement è Telcos can use AI-driven tools to better understand and analyze customer data and unlock upsell opportunities based on personalized messaging and precise marketing. This helps telcos digitize and streamline their sales & marketing operations, improve customer care and lower churn, and boost operational performance.
- Fraud detection and security è AI tools can analyze data patterns in real time, detect abnormal behaviors, and flag potential fraud or security breaches. Once identified, AI-tools can help proactively respond to threats, safeguarding an operator’s network, critical infrastructure, and customer data, thus boosting overall security management.
Telcos can use AI to totally reinvent the customer experience by creating services that revolve around extreme personalization, new immersive experiences, and improved product bundling.
Examples of telcos using AI
SK Telecom (South Korea) è In September 2023, SK Telecom announced its ‘AI Pyramid Strategy’, aiming to transform the company into an AI company by 2028 with revenues of 25 trillion South Korean won (~US$20 billion). This will be built on the back of a three-pronged strategy revolving around:
- AI Infrastructure: This is the ‘low hanging fruit’ that consists of data centers, AI semiconductors, and multi LLMs (large language models), and will serve as a technology platform that underpins SK Telecom’s transformation into an AI company.
- AI Transformation (AIX): Described as a “no regret move,” SK Telecom will use AI to innovate in its existing core business areas, such as mobile, broadband and enterprise, and expand into new business areas such as mobility, AI healthcare, media, and advertising technology.
- AI Service: To “break new ground,” SK Telecom will implement the A. strategy, the world’s first Korean LLM service. A. is a generative AI-based personal assistant service designed to help users’ in everyday life via a seamless connection to a wide variety of AI services, including SK Telecom services such as music streaming and e-commerce. A. users can create and customize an AI avatar that reflects their personality, communicate via the avatar through conversation or text messages, and request a diverse range of information.
Elisa (Finland) è Elisa Finland has launched an AI-powered chat bot called Annika that has helped it to fully automate around 70% of inbound contacts, with a 42% FCR (first call resolution) level. Annika has increased its NPS (net promotor score) from 30 to 50, above the average level for human customer services.
Telenor (Norway) è In January 2019, Telenor launched the AI-powered virtual agent Telmi. From launch to October 2023, Telmi had handled more than 1 million customer inquiries. Telmi delivers near-instant assistance at far greater speeds than traditional customer service channels, and can handle around 2,000 different predicted customer intents (real ‘wants’ vs what customers ‘say’). The clear business and ROI goals that the company defined at launch were easily achieved in the first year, and Telmi has proved itself as an additional sales channel rather than just a simple answer bot.
Zain (Kuwait) è Zain launched the AI-bot zBot as a fully automated, interactive digital channel for smart customer service. It responds without requiring input from a human agent, and can address questions such as showing the latest offers and promotions on prepaid and postpaid plans; managing accounts; adding or removing roaming services and international calls; and showing current balances or contract details.
Airtel (India) è Bharti Airtel has developed an AI-based solution to proactively detect, prevent and eliminate phishing spam through messaging. The solution creates an anti-spam filter, which, at its peak, succeeded in detecting and blocking two million messages a day. The telco piloted the solution with HDFC Bank, with direct connectivity established between the two partners to help secure E2E SMS connectivity and flag any fraudulent routes.
Vodafone (European footprint) è Vodafone is improving decision-making on network planning with software that can process and analyze up to eight billion points of data daily across its mobile network in 11 European countries. This platform, United Performance Management (UPM), performs analysis using AI tools from Google Cloud and other related services. By using real-time insights from UPM to simplify core operations, Vodafone has already seen a 70% reduction in major network and IT incidents. This frees up staff to work on other key tasks.
Verizon (USA) è Verizon uses AI tools to proactively track changes in traffic flows across its mobile networks in near real-time. This helps the operator identify network pain points, anticipate usage trends, optimize network investments, maintain network reliability, and maximize experience improvements during peak load traffic times.
AT&T (USA) è AT&T has democratized AI across their organization and their efforts have delivered significant business value. For example, the operator reports that its H2O AI Cloud could reduce fraud by 80%. AT&T’s AI-as-a-Service platform links together on-premises and data cloud providers and a set of proprietary AI services that rapidly deliver AI solutions organization-wide. The company has also co-created the H2O AI Feature Store to store, update, and share the features that data scientists, developers, and engineers need to build AI models within the organization.
The above are only a few examples of prominent global telcos using AI in their operations, and the list is expected to grow rapidly in both breadth and scope.
Key issues telcos face implementing AI & ways to address them
While operators are slowly finding success with AI solutions, there are still a few hurdles they routinely face with AI, including:
1. Unstructured or incomplete data è The one thing telcos have in abundance is data – and many types of it, including customer behavior and profiles, billing information, and location details. However, if all this data in not stored in a useful way, it hinders any AI system hoping to leverage it. The common data issues telco face are:
- Fragmented: Data is collected and stored by different systems, without a single unified database from where it can be accessed by the AI system.
- Unstructured: Data is stored as a big mass of uncategorized data without any context or explanation of what it relates to, which is not very useful to an AI algorithm.
- Incomplete: Oftentimes, telcos miss some data, and using data with missing components can lead to inconsistent or faulty learning by the AI system.
2. Need for additional technical expertise è Telcos usually have limited AI talent in their ranks. Building an in-house AI team can take a significant amount of time but yield little results
3. Technical integration è Difficulty integrating new AI tools into old legacy systems is one of the most common reasons for AI integration failures.
A good way to get around these issues is for telcos is to give AI a go, but not alone or all at once.
They must:
- First choose specific parts of their business on which to focus their AI efforts.
- Invest in fostering an ecosystem and entering into mutually beneficial partnerships with organizations that have the suitable AI capabilities.
Following this strategy can help telcos unlock the true power of AI, and leverage it to optimize their operations and ideally increase revenues and decrease costs. The era of AI has truly begun, and only telcos that harness the full potential of AI will be able to thrive in the future.
Disclaimer: Any views and/or opinions expressed in this post by individual authors or contributors are their personal views and/or opinions and do not necessarily reflect the views and/or opinions of Huawei Technologies.
Leave a Comment