By Cath Everett

August 14, 2023

Although the increased adoption of AI is unlikely to bring about the demise of the traditional IT department any time soon, it could result in a change of focus and responsibilities over time if pioneering employers like BT Group are anything to go by.

The UK’s largest telecommunications and network provider has been an AI user for many years, employing the technology in a mainly operational setting to enhance its internal systems and services. But just over two years ago, former Group CIO of Indian telco Bharti Airtel Harmeen Mehta was appointed as the company’s Chief Digital and Innovation Officer to lead its digital transformation initiative.

Heading up its Digital technology unit, which was created in April 2021, she is responsible for IT, digital innovation, BT-wide business transformation and the firm’s data and product strategy, which includes AI. As a sign of just how highly rated her role is within the organization, Mehta reports directly to Chief Executive Philip Jansen and is a member of the Executive Committee.

The aim of the Group’s transformation efforts, meanwhile, was to tap into its vast stores of data and apply AI to it in order to develop new commercial offerings. As part of the process, the supplier digitalized and centralized its business workflows, which enable activities like order fulfilment, and combined the resultant information with data from the company’s nationwide network.

Making data more accessible

This network is the responsibility of Chief Technology Officer Howard Watson who works closely with Mehta. He is accountable for network strategy, transformation and service platforms as well as cyber- and information security. As Detlef Nauck, the provider’s Head of AI and Data Science Research, points out:

We needed to pull everything together as it was important that the data from both the network and the workflows was available to everyone. So Digital’s aim was to move away from stovepipes and make data accessible to the rest of the organization so it could be used more effectively.

As a result, it introduced a hub and spoke model and appointed Adrian Joseph as Managing Director of the Group AI and Data Solutions unit that makes it work. This sits under Mehta’s central Strategy & Transformation unit and includes a cross-functional Center of Enablement - rather than Excellence as “We wanted to make it clear its remit was business-wide and it’s role was to help turn the handle on AI”, Nauck explains.

Among other things, the Center provides enterprise-wide rules for data and AI usage to dedicated teams within each of the firm’s customer-facing units. It also steers the supplier’s data migration program, which involves moving data from silos to centralized data stores.

The pros and cons of Chief AI Officers

But interestingly, the Digital organization has chosen not to hire a Chief AI Officer (CAIO) to help with the transformation. Nauck explains that this is because the unit’s remit is much wider than just AI or data:

It’s not like we have one central AI unit – that’s not how it works here. AI is a tool that we put to use in areas where we believe we can derive value – it’s not a panacea. So rather than one role, we’ve created diverse roles and created teams around them. What’s appropriate depends on each organization and how complex and diversified it is. But most companies are still stuck in a world of technology debt and sorting out data – and that’s the most important thing to do as AI doesn’t work without it.

Nauck also refers to a Harvard Business Review article entitled ‘Why Chief Data and AI Officers are set up to fail’. As he points out:

It’s because people expect too much of one individual – CAIOs have to sort out data governance, risk, skills across the organization etc. So having a distributed capability makes more sense. Also the whole space is developing so quickly that you have to learn to live with uncertainty and manage it, be nimble and react to new developments – and the kind of structure we have is well placed to deal with that.

A shift towards specialization

One trend that Nauck is seeing in skills terms as a result though is a move towards specialization. For example, he is seeing classic data scientist roles start to focus more on modelling and developing an AI solution that provides business value, while operationalizing it is undertaken by a MLOps specialist.

The influence of data and AI is also spreading more widely across the business and creating new roles there too. For instance, data protection, privacy and ethics experts have all started working in the legal department. The corporate policy department now looks after AI governance and is helping to shape the discussion around AI and data regulation. A suite of roles have also been created in corporate risk, while AI has been added to the risk register. As Nauck says:

Everyone’s looking for someone to look after and manage data and models and influence the business, but those Unicorn jobs are now breaking up into separate roles. That’s important as you need a lot of different roles and diverse voices to help shape this space. Another issue is that you often can’t hire people with these skills. It’s a developing and dynamic area so people are having to evolve with it – and you can help them do that by building up your corporate in-house training programs.

As for the impact of generative AI, Nauck believes it will be a “game-changer”:

Previous forms of AI were handled by specialists, for instance, to drive the efficiency of internal workflows and systems. But generative AI will touch everyone. So you have to think about uplifting the entire organization in terms of skills and how to use it responsibly. As a result, we’re working on a corporate-wide training programme to help people get value out of these new tools and capabilities. But there always has to be a human in the loop. With generative AI, it’s necessary to check outputs as you can’t use it in an automated fashion yet, which means people need to be trained for that. It’ll be a massive change for organizations and we currently have no idea what opportunities and risks it’ll bring. It’s so new that we just need to get our heads around it.

Taking a more incremental approach

For most organizations though, a less major revamp and more incremental approach to change is likely to be required at this point. It is also worth bearing in mind that there is “no right or wrong way to do it”, says Kriti Sharma, Chief Product Officer of Legal Tech at Thomson Reuters and Founder of AI for Good, a social enterprise that develops tech for social good:

It really depends on the business. Some companies opt for CAIOs and others for AI Product Officers. Some create a central unit with a dedicated budget, while others embed AI in each team or a hybrid of the two. Those are the main approaches that I’m observing at the moment.

Moreover, Sharma does not believe that the IT department is likely to be subsumed beneath, or replaced by, a dedicated AI function any time soon. As she points out:

Not everything is AI. There are other platforms too like engineering and infrastructure. So tasks may change and more AI capabilities will be infused into them over time but these roles won’t all disappear. In fact, the board will expect tech leaders to develop their own strategies and deliver outcomes in this area, so they should actually benefit from more empowerment and ownership to design new experiences and improve productivity. This means it’s a really exciting time of change, transformation and skills-upgrading for tech leaders, and it’s also a great time to drive more impact.

My take

AI in all its forms will undoubtedly over time change roles and responsibilities within the average IT department. But for the most part, such change will be incremental rather than turning the function on its head – although recommendations are that it should definitely be planned for rather than just winging it and hoping for the best.