Gen AI: Where should telcos start?

The key Gen AI concepts

ChatGPT defines Generative AI as:

“a class of artificial intelligence models and algorithms that have the capability to generate new, original content. Unlike traditional AI models that operate based on predefined rules or patterns, generative AI models can produce novel data that resembles the patterns observed in the training data. Generative AI models can generate content in various domains, such as natural language, images, music, and more”.

Enter your details below to download an extract of the report

The diagram below describes the range of models, concepts and uses that are seen in discussions around Gen AI.

• Blue boxes describe the main models, architectures and concepts that underpin various Gen AI capabilities (e.g., large language models).

• Orange boxes describe the general capabilities of these models (e.g., natural language generation).

• Grey arrows show the main models used to create capabilities in an orange box, and smaller black arrows show where other models can also be used (e.g., diffusion models provide image generation capabilities).

• Red text gives some of the uses made of the capabilities shown in the yellow boxes (e.g., generation of novel text).

• Red boxes highlight some of the popular foundational models for these uses (e.g., ChatGPT).

Concepts in Gen AI

Source: Charlotte Patrick Consult

Definitions of terms in the graphic:

Generative models create something new based on examples they are given.

Foundational models introduce a significant breakthrough, a new architecture or a novel approach that paves the way for subsequent advancements in the field.

Parallel dataset is a data set which provides exact translations of all words in one language to the other.

Discriminative model is a type of machine learning or statistical model that classifies input data points into different categories or classes.

GPT (Generative Pre-trained Transformer) is a foundational model which can generate text responses.

LaMDA is a Google project to provide a language model designed to allow more free-flowing conversations.

dALLe is an OpenAI system that creates realistic images and art from natural language.

Whisper is an automatic speech recognition system with improved ability to understand accents, technical language and background noise.

BERT (Bidirectional Encoder Representations from Transformers) is a transformer-based language model for text generation.

PaLM is a Google model that works on advanced reasoning tasks including code, mathematical problems, classification and question answering.

Table of contents

  • Executive Summary
    • After the hype of Generative AI (Gen AI)
    • The most compelling uses for Gen AI in telcos
    • How to scope a Gen AI project
    • 4 core recommendations
    • Next steps
  • Introduction
  • What does Gen AI bring to the telco?
  • Gen AI use cases in a telco
    • 1. Content creation
    • 2. Human-machine interactions
    • 3. Human-human interactions
    • 4. Knowledge management
    • 5. Process improvements
    • 6. Data management
    • 7. Intelligence improvements
  • Where is the value in Gen AI?
    • Important types of Gen AI
    • Use cases for Gen AI
  • Conclusion
  • Index

Related research

Enter your details below to download an extract of the report

Generative AI and beyond: Preparing for future A3

Generative AI and other technology changes

Previous work in 2020 (the basis of our report, Telco A3: Skilling up for the long term published in January 2021) uncovered four areas of A3 impact that will shape a telco into the mid and longer term. Since then, new internal and external consequences have emerged from both the telco’s and its customers’ adoption of A3, as well as changes around the underpinning technology that a telco will need to deploy – in addition to A3-induced shifts in organisational shape and focus.

 Four main areas of A3 impact

Source: Charlotte Patrick Consult, STL Partners

The figure below details the main A3 activities inside these four areas, shown against an approximate timeline which stretches from the short term into the longer term. This report addresses these activities, including thing as customer, decision intelligence, generative AI and digital immunity (as shown in the red boxes in the figure below), which we pay particular attention to due to the current high interest in the area and/or the significance of their expected future impact.

A3 activity areas for telcos

Source: Charlotte Patrick Consult, STL Partners

Enter your details below to download an extract of the report

Augmented customer experience

New A3 is used to provide support for unassisted (digital) and assisted (human agent) interactions between the telco, the telco’s customers or ecosystem partners and the telco’s supplier or partners. The figure below shows the increasing complexity of these interactions: the grey dashed lines show current interactions which are mostly human-to-human; the coloured lines show new machine interactions either for care purposes (orange) or for purchasing (red).

Entities in the new customer ecosystem

Source: Charlotte Patrick Consult, STL Partners

The newest area for telcos is the introduction of interacting with a “thing”. This is defined as a piece of user equipment (typically, a connected device or sensor or even a bot) that can interact with the telco to request care or make a purchase. The figure above shows the other entities within the environment.

  • Centralised purchasing bot: Designed to purchase goods and services on behalf of a company or individual.
  • Embedded intelligence: Intelligence added into a thing which takes it from being able to make simple requests (“I need help”) towards being able to collect data from multiple sources and create more sophisticated requests (the infamous smart refrigerator ordering groceries). Embedded intelligence in the telco network may also be able to receive more complex requests and prescribe/execute remedies in downstream systems.
  • General consumer bot: Amazon Alexa, for example.
  • Contact centre botand sales bot: These interact with humans or machines to provide help or take an order.

 

Table of Contents

  • Executive Summary
    • Developing A3 will significantly impact telcos in four areas
    • Preparatory actions for telcos
    • Activity streams: A summary
  • Introduction
  • Augmented customer experience
    • Main concepts
    • Thing as customer: The significance for telcos
    • Next steps for telcos in augmented customer experience
  • Augmented experts
    • Main concepts
    • Decision intelligence: The significance for telcos
    • The next steps for telcos in augmented experts
  • Intelligent automation
    • Main concepts
  • AI design
    • Main concepts
    • Generative AI: The significance for telcos
    • The next steps for telcos in AI design
  • Smarter customers
    • Main concepts
    • The next steps for telcos in supporting smarter customers
  • Increasing intelligence
    • Main concepts
    • The next steps for telcos in increasing intelligence
  • Trust, value generation and skills
    • Main concepts – trust
    • Main concepts – value generation
    • Main concepts – skills
    • Digital immunity: The significance for telcos
    • The next steps for telcos in trust, value generation and skills
  • Conclusion
  • Index

Enter your details below to download an extract of the report