The AI Edge In Telecom
Analyzing the business impact of Generative AI
Sponsored by:
Hundreds of companies are collecting and trading your personal data
Get your data out of data brokers’ hands. Stop robocalls, spam, stalkers, & identity theft now. Keep your information secure with Incogni. Just for readers: Get 60% off Incogni using code NL2023.
GenAI Overview
The fastest-growing advanced AI technology today is generative AI.
This research centers on generative artificial intelligence (AI) and offers an analysis of the opportunities and sector-specific business impact associated with the telecom market segments.
In the coming years, generative AI will likely pose a threat to businesses in every industry.
As generative AI improves in precision and capability to offer dependable factual guidance, its influence will transcend industries and business operations.
Use cases include code writing, product design, and customer support chatbots, as well as the creation of news and marketing content.
Enterprise management and AI implementation
The appeal of generative AI is that it has the potential to complete tasks which current AI programs are unable to do. Writing code, creating training data and generating text can lead to a variety of applications, both in general and specific to certain industries.
Enterprises need to determine which model is best for them and what level of customization is necessary for it to work.
AI and ethics teams should be set up to make sure new AI use cases are in line with ethical standards.
Human oversight should be included in processes involving generative AI and look for solutions that explain the sources of the output.
Nations are looking into the potential risks generative AI can bring to privacy and organizations need to adhere to laws and regulations on a global scale.
Important Considerations for Assessing AI Solutions
Brand Reputation: In light of the widespread doubt and controversy surrounding the veracity of generative AI, reliable technology partners are already making it accessible through enterprise-level platforms and tools. Popular brands are responding to customer demands for solutions with built-in safety features and guardrails.
Security strength: Enterprises are concerned that their proprietary data will inadvertently end up in the public domain or be used to train large language models, making it accessible to others. Solution providers must ensure the security of enterprise data and customer information.
Strength of the ecosystem: A dependable and unified partner program, supported by prominent cloud platform providers, global SIs, and SPs, enables foundational alliances between reputable technology providers (as previously stated) and cutting-edge AI services developed by start-ups or pure plays with significant innovations. The integration of generative AI into an organization's current IT infrastructure and the provision of a seamless, end-to-end user experience will facilitate its adoption.
Explainability: Ensuring the explainability of AI, including the ability to discern the origins of generative AI content, is critical for organizations to have faith in model outcomes and act appropriately on discoveries.
Advanced data architecture: Involved data professionals in the deployment of advanced analytics projects benefit from a more streamlined and consistent experience due to the data management capabilities, including data fabrics and other solutions, which eliminate data silos and streamline information access across data repositories.
Professional services and go-to-market strategic planning: For new users, a vendor with a solid reputation among reference customers and pricing models that accommodate flexible consumption are indispensable. Deploying a generative AI solution, however, can be a challenging endeavor, even for businesses with substantial internal resources. Decision-makers ought to seek out service providers who offer the assistance of a team of specialists throughout the procedure.
Performance: Organizations that adopt generative AI as the market develops will seek solutions that exhibit consistent low latency and reliability, generate a wide range of outcomes, effectively maintain meaning and intent, safeguard enterprise data, and provide filters that function as guardrails.
AI Market Forecast
Advances in machine learning (ML), driven by enhanced algorithms and augmented computing capabilities, have enabled AI to address practical, real-world issues.
It is projected that by 2030, the overall AI market will reach a valuation of approximately $908.7 billion.
Specifically, the market for specialized AI applications is expected to grow significantly, reaching an estimated value of $477.6 billion in 2030, a substantial increase from its $31 billion valuation in 2022.
In the initial stages, investments in AI are anticipated to be primarily focused on areas such as computer vision and conversational platforms.
Telecom + GenAI
Impact on the telecommunications sector
Personalized experiences: For many years, telcos have taken advantage of chatbots, which provide automated answers to questions and complaints from customers. However, those using generative AI have a better ability to personalize customer interactions and outcomes, improving their experience. Generative AI is also a great asset to telcos in the e-commerce space, as it can aid customers in selecting the best service plans and devices for them, ultimately reducing customer churn and optimizing the customer lifecycle.
Streamlined Operations: Generative AI has the potential to increase efficiency and productivity for telecommunications companies by streamlining their corporate center and field service operations through automation. Applying the capabilities of Machine Learning and Natural Language Processing, Generative AI can automate software development processes by generating code and troubleshooting via text or voice. Furthermore, it can be used on field service devices to speed up network diagnostics, analysis, installation, and troubleshooting.
Autonomous Network: Generative AI can pave the way for autonomous telecommunications networks by connecting large and complex generative AI models to enable network planning and design, network optimization, spectrum management, load balancing and resource allocation, network slicing, and the self-healing of networks. Generative AI models can be used to analyze historical patterns of usage metrics and network performance metrics to choose practical topologies for expansion. They can also be used to automatically allocate resources, manage spectrum allocation, perform network slicing to enable near-autonomous networks, and more.
Potential Use cases
Customer support
Generative AI-driven chatbots/virtual assistants have the potential to transform customer service delivery channels for telcos, allowing for rapid recognition of customer queries and the provision of automated yet human-like conversational replies with shorter wait times.
Generative AI chatbots don't just respond to customer inquiries regarding products, services, billing, or technical issues. Also suggest and promote personalized products and services to customers based on previous conversations, leading to a better customer experience.
By combining generative AI with traditional customer service channels, telcos can offer translation services to customers who have different native languages, allowing them to reach out to customers from different parts of the world while saving money.
Operations
Generative AI can be used to take away the need for employees to carry out mundane, repetitive tasks, allowing them to channel their efforts into more important and beneficial activities. For instance, Generative AI can be used to automate software creation processes such as generating code and testing, which can lead to increased efficiency and cost savings.
Generative AI can be used to accelerate the diagnosis and assessment of field service machinery, as well as assisting in installation and problem solving.
Translating documents from the language spoken by customers and staff into other languages can make the paperwork clearer and simpler to understand.
Network Performance
GenAI can be of help to telecom companies in locating network problems by studying network traffic and providing appropriate steps to fix them, such as making sure more effective routing of traffic or increasing capacity to congested areas of the network.
GenAI can recognize patterns in network traffic and data from network devices, making it possible to foresee network problems and hardware malfunctions before they happen. This offers telcos the advantage of being able to take corrective action or plan preventive maintenance, leading to fewer network disruptions.
GenAI can also consider population density, customer demand, and other elements to suggest appropriate places for the installation of new cell towers and other infrastructure resources.
Sales and Marketing
GenAI models can aid telcos in segmenting customers efficiently and pinpointing target markets by processing enormous amounts of demographic and consumer information.
Telcos can use the spending habits and purchase history of customers to forecast their preferences, allowing them to create customized marketing materials and suggest products, services, or special offers.
These models enable telcos to develop attractive content to interact with their customers, and to supply automated sales help, as well as generating possibilities for the sale of additional or related products.
Real-world use cases of GenAI in Telecom
SK Telecom and e& have been tapping the power of generative AI to disrupt their traditional customer service/support channels and improve overall customer experience
SK Telecom
SK Telecom, in collaboration with Singtel, E& and Deutsche Telekom, declared the formation of the Global Telco AI Alliance that will construct a tailor-made Language and Learning Model (LLM) for digital assistants employed in customer service.
SK Telecom has leveraged generative AI capabilities for its A dot (‘A.’) super app, through which users can directly connect to various third-party services like music, video, e-commerce, and navigation.
Embedded with advanced NLP and sentiment analysis technologies, ‘A.’ allows customers to create and communicate with an AI character
SKT also added ‘Chat T’ feature to ‘A.’ app using the ChatGPT model of Microsoft’s Azure's OpenAI service. Users can pose their questions to ‘Chat T’ and receive rich conversational responses instead of routine answers.
SK recently invested $100m in Anthropic, and acquired 20% stake in Konan Technology to develop industry-specific LLM models.
“With the massive overhaul of ‘A.’, we expect more customers to feel comfortable and enjoy using our conversational AI,”
“We will continue to evolve ‘A.’ into a service that can help customers in their daily lives in all kinds of ways.”
Kim Yong-hun, VP and Head of AI Service Business Office, SK Telecom
E& (Etisalat)
E& has integrated generative AI models into its GoChat Messenger app, allowing users to create unique content in the form of text, audio, and video, which can then be distributed to their connections. Furthermore, these models can give precise and interesting answers to inquiries from customers inside the app.
Generative AI algorithms are being used to examine large amounts of data to draw out valuable analysis to create individualized and specific marketing plans for various types of customers.
GenAI models help employees boost their effectiveness and output by furnishing precise and prompt replies to customer inquiries, thus enabling them to center their attention on other imaginative and productive activities.
Telecos are using generative AI to enhance the performance of their voice and video call applications. Through the use of generative AI models, users can generate videos and create more engaging content, thereby increasing the quality of their experience.
In August 2023, E& invested $25 million in Ikigai Labs, a startup dedicated to the utilization of generative AI for tabular data.
AT&T
AT&T is employing generative AI to improve internal efficiencies and optimize resource utilization
AT&T deploys generative AI algorithms to analyze large datasets to extract insights and deliver quicker and more accurate answers to customer queries.
These AI models help translate employee and customer documentation from English to multiple languages, thereby eliminating the need for multilingual experts. Also, genAI-powered chatbots answer queries accurately in multiple languages, thereby reducing the costs associated with hiring language experts.
AT&T is exploring generative AI to automate repetitive and rule-based tasks across different departments, such as billing, invoicing, and inventory management. It has developed Ask AT&T, a generative AI based tool for accelerating coding and software development
It is also leveraging generative AI algorithms to optimize resource allocation, such as workforce scheduling and inventory management. This ensures that resources are efficiently used, reducing waste and operational costs.
Beeline
Beeline Kazakhstan, the operating entity of VEON Ltd., developed AI language model called BeeBERT to bolster its digital services.
The generative AI technology is embedded into the company’s website and mobile app to support customer service and enhance their experience.
BeeBERT is also being extensively used to support customers where Kazakh language comprehension is required. The AI model has exhibited increased quality in recognizing queries asked by customers.
The company is also exploring the possibility of developing generative AI models that can be trained in local languages and can give contextual answers in local languages.
“The development of BeeBERT underlines not only our technological know-how, but also our dedication to use technology to advance local development.”
-Evgeniy Nastradin, CEO of Beeline Kazakhstan
Key challenges for telecom sector in using Generative AI
Cost
Generative AI costs a lot, so acquiring the infrastructure and updating it must be taken into account. Training AI models can be costly, as they require a lot of computing power and expensive NVIDIA GPUs for queries, which cost around $10,000 each. For example, telcos would need hundreds of these GPUs to train their models.
Data Quality
Implementing generative AI requires a huge volume of data to learn from, but accuracy of AI models and their output depend on the quality of data.
Telcos must ensure generative AI models get high-quality and most granular customer and operational data.
Telcos could be restricted from accessing such data for several reasons, such as data privacy regulations or technical limitations of capturing and managing such data.
Data Security
Since huge volumes of data, including customer, operations and network traffic data are used for developing and training LLMs that underpin the use of generative AI by telcos, guaranteeing security of such data is a top priority.
AI could be vulnerable to manipulation and security breaches. Any leakage in telcos' customer or operational data from LLMs could put not just their systems and intellectual property at risk but also their reputation.
It is critical that telcos have better governance of data and AI to ensure that data privacy and security are guaranteed.
Regulatory
Telecom industry is heavily regulated, which can make implementing generative AI costly and challenging.
Many telcos operate in multiple countries with their own data protection and AI regulations. This can be complex and challenging for generative AI systems.