The Foundations for AI at Scale

Posted by Shammah Banerjee | 05-Mar-2020 18:09:27

Advancements in artificial intelligence has left organisations striving more than ever for digital transformation success. With the business potential that can be unlocked with AI, it is crucial to lay the right foundations to innovate in the next decade.

However, research shows that 81% of executives aspire to derive value from AI, but don’t understand the data and infrastructure required to support it. As this year’s Digital Trends Report showed, cost of the technology proved to be a big hindrance for implementation with 45% struggling, as well as lack of skills to enable the change (42%).

The journey to AI at scale is clearly a bumpy one from a data, skills and cost perspective. Nimbus Ninety members gathered in Prince Philip House opposite St. James’ Park to discuss how to navigate this path and build the right foundations for leveraging AI at scale.


When we look at AI implementation in businesses, the stats speak for themselves. 85% view AI as a strategic priority, and yet 51% find operationalising, sustaining and scaling AI is a major challenge. But what are these challenges?

The universal starting point is the data. As IBM’s Executive Partner and Cognitive AI Lead, Clare Mortimer reminded us, there is no AI without IA (information architecture). For many organisations, they don’t entirely know where the data is that can help build a real transactional system. To build this from the ground up requires starting with the absolute fundamentals. Collecting and organising the data is certainly time-consuming and takes a while to get off the ground, but getting this right allows the organisation to drive something very stable.

For many organisations, the drive to get the most out of data is to improve customer experience. Indeed, for 32% of organisations in 2020, AI deployment is set to impact interaction with the customer. However, the problem is that customer experience often doesn’t get better because of AI but rather is stalled as data must be prepared before the AI can truly reach its potential.

Clare gave the example of Woodside, an Australian energy company. With the centre of their workforce being 50-something-year-old engineers, the organisation faced a problem of an enormous amount of knowledge retiring from their organisation. IBM stepped in to integrate AI throughout the organisation and the result was a renewed access to that knowledge – through AI and data organisation.

Once this data is suitably collected and organised, then analysis of the data and finally, infusion of the AI throughout the organisation can take place. This can fundamentally change how the organisation works, and thus take it on a journey to modernisation.

Trust is another major challenge for organisations implementing and using AI. In a world where trust matters hugely for consumers and employees, every organisation needs transparency for how their AI is making decisions. Getting people on board in this sense is central to enabling them to respond to the decision that the AI has made, and thus, truly infusing the technology through the organisation. Often a mindset shift is required to start using the technology and realise the difference it makes.

As we saw earlier, lack of skills within a workforce also has a severe impact an organisation’s ability to implement and scale AI. “Skills are needed to build the infrastructure, that’s a given. But skills within the organisation to adopt this computer-driven system is needed too,” explained Clare. For many organisations, a skills transformation will need to happen in parallel to a digital one – and it won’t just be the data science team that this impacts, but a cross-functional effort.



For many of the members, embedding a data-driven culture with the organisation is the first step – and this requires a skills transfer to undertake the IA side of AI. Similarly, the data needed requires strict disciplines in terms of achieving the necessary quality for AI excellence. With this data-first stance, new business models can emerge to disrupt the current way of working, thinking and engaging with stakeholders to drive more than just modernisation.

Culture came up again and again as a big area for organisations to address when it comes to AI implementation and scaling. It isn’t enough to bring the new technology in: a mindset shift is needed to truly bring the people in the organisation along with you. As discussed above, this is partly to do with trust of the technology and how it is making its decisions, but also to do with truly understanding the potential of the technology and being led through that change. Engaging stakeholders at all levels is vital to fulfil a technology’s potential and see real ROI.

Iterating the technology again and again was also cited as a way to positively develop the technology’s impact on an organisation, as well as driving innovation further. AI isn’t just the end goal: it is the enabler to greater innovation.

The discussion around how to achieve an ethical approach to data gave many different viewpoints on the best methodology: one member suggested eliminating the black box, while another suggested that accountability is needed to balance privacy with AI norms. Overall, it was agreed that ethical approach really differentiates across organisations, but frameworks are necessary to keep a base level of ethical practice across industries.

While it unlocks huge business potential, there is no question that implementing AI poses a huge challenge to businesses – both in terms of resources and mindset. Organisations will need to plot their course forward, beginning with data, before embarking on the journey to AI at scale.


This event held was in partnership with IBM, an IT solutions and cloud provider.


Topics: Event reports

Written by Shammah Banerjee

Shammah is the Senior Editor at Nimbus Ninety. She tracks down the most exciting stories in business and tech, produces the content and gets to chat with the biggest innovators of the moment at Chief Disruptor LIVE.

Leave a Comment