Generative AI (GenAI), as part of AI and machine learning (AI/ML), took center stage in the telecoms industry in 2024. A lot of operators and CSPs experimented heavily in 2023 and 2024 to select which proofs of concept (POC) have achieved the highest merit to deliver tangible outcomes.
Cutting through the hype and focusing on the right questions, operators identified how to tackle the emerging generative AI landscape and what additional value the new technology can bring to their overall data and AI journey. While natural interaction through large language models (LLMs) has greatly appealed to generative AI users, the additional value is delivered by bolstering the field with new use cases.
While traditional AI/ML requires structured data sets of high quality to train efficient models for often very specific tasks, generative AI removes barriers to additional use cases. As foundation models can be used with unstructured data, telecom operators can tap into their large pool of data assets to monetize a broad range of new use cases and select those which provide them with the shortest path to a maximized return on investment (ROI). This has led to a large number of evaluations, exposing commonality in the industry, while also ensuring operators retain the freedom to select a starting point specific to their unique AI and cloud journey.
Successful use cases, based on real experimentation, proof of concepts, and proof of value (POV) can largely be clustered into four areas:
- Customer Engagement: Cloud contact centers, powered by AI, can improve agent productivity and customer experience, as well as forecast the needs of end customers to suggest retention offers or personalize recommendations.
- Business Operations: Encompasses automation in sales RFx responses, automating code generation, debugging, testing for system deployments, and the discovery of revenue leakage.
- Specific to Telecom: Network operations can be enhanced with smart planning, installation, configuration, and optimization. This further extends to troubleshooting and the automated healing of networks.
- Identification of New Revenue Sources: Here, the scope depends on the CSP’s actual business but success has been seen in leveraging an enhanced smart home (IoT) experience, services that utilize edge inference and analytics, the hyper-personalization of offerings through a connected end customer journey, and security services that address anti-fraud and user authentication.
Each operator has the task to identify a catalog of use cases they now need to implement in production environments. While some already started this in 2024, the next phase of generative AI deployments holds new challenges and requires the removal of new barriers. As a result, it drives a number of common questions that center around project prioritization, cost management, model selection and evolution, risk management, and compliance. These questions include: “How can we move faster?” and “How can we scale properly?”. It puts the proper design for AI, especially for generative AI, into the spotlight. It also raises the point of generative AI maturity for each operator as it relates to the following areas: technology understanding, organizational skills, data readiness, security and compliance, business objectives, and production readiness.
To help the industry navigate these new challenges and to gain the required insights, AWS has collaborated with TM Forum as a sector industry organization to develop the Generative AI Maturity Interactive Tool (GAMIT). Launched at Innovate Americas in September 2024, and based on initial data from more than 200 CxOs from telecom organizations, this tool provides the opportunity for each CSP to benchmark themselves against the industry in relation to each of the maturity areas.
At the same time, these exercises provide additional insight captured in a growing and evolving data foundation, increasing value over time. Compared to traditional research exercises, the tool provides each participant with organization-specific insights and analytics on their maturity posture for generative AI compared to the industry average and leaders in the field. With this additional information, operators can decide which focus areas to prioritize and where to look for best practices to accelerate their journey. To achieve this quickly, the individualized results match a catalogue of assets from TM Forum and AWS that provide a quick start. The tool can be accessed using the QR code published in this article.
In addition to maturity and readiness for AI/ML and generative AI, there are a number of design considerations that are important, including the selection and choice of a respective infrastructure, an integrated data pipeline, plug-and-play AI/ML models, and strong security and governance.
To ensure the seamless integration of generative AI with traditional AI/ML and the wider landscape of IT and operations systems, access to a broad partner ecosystem and an intuitive builder experience for everyone involved in the deployment of the use cases is required.
The new imperative is for telcos to become cloud and AI-first organizations, with the ambition to AI-power everything, everywhere, requiring them to carefully select where to start and how to prioritize. CSPs that are starting with generative AI must select the right use cases and empower their teams to innovate with tools and training. Again, the fastest path to return on investment (ROI) relies on utilizing top use case selections that will result in tangible success and measurable benefits.
In 2025, operators will have to prove long-term ROI and demonstrate the value of their generative AI production deployments prior to implementation. The aim is to create a flywheel effect that helps scale the return on the necessary investments in the platforms, skills, and partnerships, while optimizing costs and keeping output performance, compliance, and quality high.
In terms of multi-modal models, evaluating more efficient purpose models, such as small language models (SLM) and domain-specific models; creating effective MLOps; and optimizing AI infrastructure, are all aspects that decision makers in CSP organizations must keep in mind while accelerating their AI/ML and generative AI journey in the next two-to-three years.
The key factors for success include being able to move quickly, effectively customize and scale, and make the flywheel spin.
Also Read: AWS Announces Data Center Innovations for 2025
Also Read: AWS Launches Generative AI Partner Innovation Alliance
