Until around 2014, most cutting-edge artificial intelligence (AI) research was conducted by academic institutions. However, post-2014, commercial organisations started investing in this field. In the near future, AI and AI-based applications will take centre stage across various sectors, including agriculture, pharma, defence, automobile, environment and sustainability, healthcare, entertainment, information technology (IT) and telecom. AI is also rapidly becoming consumer-facing.

AI deployment promises to change how the world operates, but also presents a new set of challenges and dangers. It can be used to bypass older cybersecurity mea­sures, rendering tools like voice recognition insecure as it can easily clone voices. It can be used to cause harm to common citizens – with the misuse of face recognition, for instance – which can easily compromise privacy.

In the telecom industry, AI deployment in networks for operations, customer service provisioning, delivery management and infrastructure operations has already led to significant cost savings. It has also enabled faster roll-outs, improved network reliability and enhanced customer experiences. It has reduced both time and costs associated with repair and maintenance, leading to an improvement in the quality of services delivered.

Furthermore, telecom service provide­rs can explore new revenue streams throu­gh the use of AI and big data, especially with the roll-out of 5G networks (and later 6G networks), which will enable a significant scale-up in technologies such as internet of things (IoT). They can offer new suites of AI-driven services to enterprises, leveraging the rapid convergence of IT operational technology and IoT. Nasscom estimates that AI can potentially add $450 billion-$500 billion to India’s GDP by 2025, a substantial increase, given India’s current GDP of around $3.7 trillion.

Policymakers are also aware of the po­ssi­bilities. There is a clear focus on transforming India into a global AI hub, with a combination of initiatives to foster resear­ch and build semiconductor and related manufacturing capacities. Reports from the Ministry of Electronics and Informa­tion Technology (MeitY) and the TRAI, as well as industry insights, highlight the new opportunities and challenges associated with AI. India’s corporates are also investing in AI and AI-related research while seeking collaborations with academia and overseas MNCs.

The Government of India has taken concrete steps to encourage the adoption of AI responsibly and build public trust in the technology. The concept of “AI for All” is at the core of the National Strategy for AI. India has played a key role in shaping the global discourse around emerging technologies. As one of the largest eco­no­mies in the Global South leading the AI race, India has been entrusted with the responsibility of serving as a council chair for the Glo­bal Partnership on Artificial Intelli­gen­ce for a three-year tenure, the incoming chair in 2022-23, lead chair in 2023-24 and outgoing chair in 2024-25.

The government’s working groups on AI have recommended setting up a three-tier compute infrastructure comprising 24,500 graphics processing units (GPUs). A report released in Oc­tober 23 by Rajeev Chan­dra­se­kh­ar, MeitY, recommended establi­shing best-in-class AI compute in­frastructure at five loca­tions with 3,000 AI Peta floating-point operations per second of computing power. This is a 15x increase in capacity compared to the current installation.

The working groups have recommended the implementation of a three-tier model to establish India’s AI compute infrastructure for high-end compute, mid-range compute and edge compute, utilising GPU processors deployed in gaming devices or high resolution multimedia ap­p­li­cations. The groups have also suggested setting up an inference farm to host AI applications to solve real-time problems or provide solutions.

On the corporate side, GPU maker Nvidia has collaborated with Jio and the Tata Group to build the compute infrastructure with the required resources. IBM has signed three MoUs with entities engaged with MeitY to advance and accelerate innovation in AI, semiconductors and quantum technology. These MoUs will help MeitY leverage IBM’s expertise.

IBM and IndiaAI – Digital India Cor­poration plan to collaborate to establish a world-class national AI innovation platform with a focus on AI skilling, ecosystem deve­lopment and integration of ad­vanced foundation models and Gene­rative AI (GenAI) capabilities to support scientific, commercial and human-capital development.

IBM will also be a knowledge partner of the India Semiconductor Mission (ISM) for a semiconductor research centre. IBM may share its experience with ISM on intellectual property, tools, initiatives and skills development, aimed at promoting innovation in semiconductor technologies such as logic, advanced packaging and heterogeneous integration, and advanced chip design technologies, using modernised infrastructure.

IBM and the Centre for Development of Advanced Computing will also explore opportunities to contribute to the advancement of India’s National Quantum Mission by building competency in quantum computing technology, applications in areas of national interest, and a skilled workforce.

Jio is offering AI services powered by Microsoft Azure to enable enterprises to build their own in-house AI solutions on the Jio platform. The company is also offering cloud AI solutions, which can be customised for corporate needs. Bharti Airtel has started using speech recognition and speech analytics algorithms on its network to improve the quality of customer services. This initiative, built and deployed in collaboration with Nvidia, is part of Airtel’s enterprise-focused AI solutions.

At the academic level, the IIT Madras Centre for Responsible AI (CeRAI) has announced its partnership with Ericsson for joint research in the area of responsible AI. IIT Madras also hosts a key test bed for 5G. CeRAI is an interdisciplinary research centre for fundamental and applied rese­arch in responsible AI with an immediate impact in deploying AI systems in India.

Ericsson has signed an MoU with CeRAI for five years. Ericsson Research wi­ll support and participate in research acti­vities at CeRAI. AI research is a key focus area for Ericsson as 6G networks will be autonomously driven by AI algorithms.

CeRAI already has several projects on applications specific to the telecom industry. One of the projects on participatory AI addresses the black-box nature of AI. The project studies governance mechanisms that enable stakeholders to provide constructive inputs for better customisation of AI, improve accuracy and reliability, and raise objections over potential negative impacts.

GenAI models based on attention mechanisms have demonstrated exceptional performance in tasks such as machine translation, image summarisation, text generation and healthcare, but they are complex and hard to interpret. A CeRAI project on the interpretability of attention-based models explores the challenges in interpreting and understanding the patterns in data that these models learn.

A project focused on multi-agent reinforcement learning (MARL) for trade-off and conflict resolution in intent-based net­works is of special interest to telecom networks. This project studies MARL me­thods to manage coordination and facilitate automatic cooperation when intents conflict in networks.

Data centres are another segment in which AI is leading a revolution. Very soon, data centres are projected to consume a whopping 20 per cent of global power su­p­ply, and AI and ML integration will be­co­me the new normal. Data centres will need to be redesigned to carry the new co­mputing and power workloads, and manage the physical requirements of cooling. This must be done in an environmentally sustainable way.

India is one of the fastest growing data centre markets. AI and ML are required to increase energy efficiency and sustainability, improve asset performance management, enhance customer relationship ma­na­gement and security, and improve capa­city management and planning in the data centre space.

As GenAI advances, several applicati­ons are expected to emerge. It can create new content based on data patterns it was trained on. It is an advanced version of predictive AI, which relies on AI algorithms to provide forecasts or predictions based on its training. GenAI can find new uses across telecom networks.

GenAI can create and reverse engineer human-readable content. GenAI can pro­du­ce human-readable content such as texts, soft­ware code, images, music and videos fr­om set inputs, as well as produce text descriptio­ns from non-text inputs, such as an image.

These capabilities can be implemented across diverse functions such as marketing and sales, customer service, operations, legal, reporting and analytics, career development, and even in the software development life cycle. In telecom, the applications can range from the creation of service-level agreements (SLAs) and product documentation to generation of network characteristics from text-based documentation, such as customer SLAs. Intuitive dialogue-based interfaces such as ChatGPT can also be introduced for expert systems, making user interactions easier.

GenAI can produce machine-readable content for mobile network data, raw-format logs or network configuration parameters to produce coverage maps. It is also used for incident identification or detecti­on, search optimisation, configuration recommendations or even resource allocation.

In addition, GenAI can be used to synthesise additional data to augment existing data sets for training or other purposes, particularly when data is scarce or expensive to collect. In networks, this may be due to a low volume of connected devices at a particular time, technical malfunctio­ns, or an overloaded network.

GenAI can create semantic communication by encoding information into a mo­re compact, compressed format that represents raw information in a way that it can be decoded or synthesised. By avoiding the need to transmit raw data, this approach enhances transmission efficiency and saves bandwidth.

GenAI can also play a key role in developing digital twins, which are virtual representations of physical objects, processes, or systems created to simulate and model the behaviour, performance and characteristics of the physical object. This can be extended to an entire mobile network, a single network function or a wireless protocol, to test, analyse, optimise, monitor or validate a system without any risk to the actual network. GenAI can be rapidly trained on the physical network to create digital twins. GenAI also can be used for the creation of interactive virtual environments where algorithms could be trained and tested.

Telecom is one of the industries at the cutting edge when it comes to AI deployment. With AI, telecom service providers are well positioned to offer services such as cloud computing, data centres, and enterprise solutions and analytics to clients. Given the positive externalities of the telecom industry, it could be one of the key drivers for AI adoption.