Taking AI out of the shadows

If artificial intelligence (AI) has been the technology story of the last two years then digital transformation will need to be on the agenda for organisations everywhere. The recognition that technology – and customer expectations – are changing, then, can be seen as a welcome development.

In fact, said John Tallon, director of modern workplace and modern apps at Storm Technology, one of the main effects of generative AI was that it had democratised AI.

“Machine learning [ML], for analogy detection, is very technical and used by data scientists in specialist use cases, not even by AI teams. Generative AI, on the other hand, everyone can understand it and see the use cases for it.”

Even generative AI is not entirely new, in fact, though it is undeniable that interest surged after the release of ChatGPT 3.5 by OpenAI in late 2022.

“My take on it is that it wasn’t just a start-up, it was a well-funded start-up, and Microsoft vouching for it as a platform gave it a lot of credibility, but that came after its initial releases,” Tallon said.

Whatever about the potential uses for AI in business, though, the fact is that staff are already using it, albeit often in an unauthorised and ungoverned manner.

A survey conducted by Storm found significant use of AI as a form of so-called ‘shadow IT’.

“We did a survey in November-December [2023] and got feedback from 110 c-level people in Ireland. A high 70s percentage admitted AI was being used in a shadow capacity,” he said.

Awareness of how AI could be used was lower, at just 63 per cent, but there was an acknowledgement that it was inevitable: “Everyone said it would have an impact.”

The appropriate response to both the reality of AI use and its potential, Tallon said, was to develop an AI strategy that looked at both business cases for it and governance.

“There is now a focus on putting together small teams that can discuss use cases, then they can select one and try it,” he said.

For example, parts of the public sector are interested in using AI to help demystify the content coming out of government.

“Can something like an LLM be applied using a retrieval augmented generation to give a short, snappy answer to a question? That’s a citizen-facing benefit,” he said.

Internal uses are also high on the agenda, he said, such as summarising the blizzard of documents businesses and public sector bodies create and have to deal with.

“Also, there are more advanced options being considered, such as in scanning and systematising documents. Generative AI has sparked off applications for AI that already exist, such as OCR and speech-to- text. Frankly, it has rebooted the whole machine learning and AI field.”

In recent conversations with executives data is a big focus

While there are implications for every sector of industry, naturally there is variation. For example, construction is particularly hard to digitise, Tallon said.

“You’re dealing with the material world and the supply chain, and of course weather. Retail, on the other hand, needs to be super-competitive due to the small margins, but has great data.”

Internally within organisations, be they in the public or private sector, conversations about AI need to be part of a wider digital transformation conversation encompassing spend, culture and change management.

There is some recognition of this, said Tallon, but data governance, including cleansing, needs to be taken seriously.

“In recent conversations with executives, [we have found that] data is a big focus. But you need data prep for that. There is work to do,” he said.