Sulabh Soral, Chief AI Officer for Deloitte’s UK arm, tells Martin Cooper MBCS about the steps that can help businesses transform AI from an idea into a technology that delivers value across an enterprise.
Sulabh Soral pauses, says: ‘That’s a good question’, pauses again and then smiles broadly. ‘How do I think historians will view this period in AI’s development? In the moment, technologies can look negative. When history books are written, technology is looked at in a very different fashion. And I think historians will see this as the single most important thing humans have done to augment civilisation. That’s how I see it. This is a huge moment of uplift.’
Sulabh is Chief AI Officer for Deloitte’s UK arm and, he explains, he spends most of his time helping the firm’s clients transform their businesses by embracing AI. Along with being outward looking, he’s also part of a Deloitte’s Office of Generative AI that looks at how the global consulting firm can transform and advance itself by using the technology. Part of that second role sees him taking a lead in developing AI technical talent within the firm.
We start by discussing the question Sulabh sees as the first step on the road to AI success: how should a business interested in AI begin its relationship with the technology?
Before delving into tactical, practical and strategic advice, Sulabh says it’s essential to take a step back and understand what AI will eventually become. Today, the technology is sometimes viewed as separate or unique, but over time, he suggests that AI will become part of an enterprise’s operating fabric. AI will become as ubiquitous and as embedded as the cloud.
Taking the first step with AI
Before we get to that point, though, businesses will need to take their first steps. Focusing on generative AI, Sulabh says: ‘The best way to start is to get used to it. Try AI, touch it… play with artificial intelligence. Generative AI is a new way to do computation — to talk to computers. It’s a new interface — that’s the most important thing to recognise.’
He explains that businesses that embrace AI successfully often begin early and with almost playful, proof-of-concept work.
‘The next thing to do is build an enabling operating environment where this AI can scale’, he says. ‘If you find something useful, how do you scale it?’
In practical terms, Sulabh explains that this means revisiting and potentially updating data, risk, legal and even training policies — all in addition to having the right technology and data infrastructure in place. ‘You really need to think about how you’ll enable your people to become productive and fluent in AI’, he says.
Don’t think small
Successful businesses that have developed to become successful AI businesses also view the technology differently. The key, Sulabh believes, is to think big. Again, in practical terms, this means viewing AI as something to be embedded in an organisation's strategy, something that becomes a platform that runs across the whole organisation and upon which many different functions are built. This differs from viewing AI simply as a tool that might generate cost efficiencies by automating discrete business processes.
‘That doesn’t deliver a return on investment’, Sulabh says. ‘Just spinning up a chatbot for better customer experience may not generate ROI. Companies that think about large scale transformations — the operating fabric story — these are the companies where you see value being generated.’
Adding weight to his point, he explains: ‘Think about process automation… no process in an enterprise is a single step; every process has many steps. What people don’t realise is that if you have a 10 step process and you automate nine steps using AI, it still doesn’t automate the process. This isn’t about replacing processes; this is about reimagining processes and updating our organisations. It’s a huge opportunity.’
Drawing an analogy, Sulabh winds the clock back to a time just before a previous IT revolution began: the arrival of the desktop PC. He asks: ‘When the PC came along, if every company had said “we’re getting PCs for our employees, let’s figure out the use case”, do you think the PC would have been adopted?’
The answer is clearly no. Rather, Sulabh explains that PCs were successful because everybody with a business could compute and that increased productivity across the organisation. Suddenly, the tax, finance, HR and product departments all got PCs.
The rule of three
There are, Sulabh explains, three components in successfully scaling an AI product or project within a business. The first is data. Generally, Sulabh says, businesses will find they’ll extract much more value when mating AI with their own data than with information found on the internet. This means organisations need to invest in organising, qualifying, sifting, securing, managing and checking data before it is presented to an AI.
Next is trust — do you and your people trust the AI’s recommendations? And finally, businesses need to build fluency in AI as a tool and that fluency needs to extend right up and across the organisation.
With trust, data and fluency established, an AI project stands a good chance of moving from a concept to a reality. The next step is thinking about how team members or any other intended users will use and adopt the final product.
‘You need to focus on accessibility’, Sulabh says. ‘Not everybody is equally comfortable with technology and, thinking about user interfaces, people may have different requirements — that could be visual or audible — we’re not all the same. The point is, if we don’t make AI accessible, people can’t use it.’
For you
Be part of something bigger, join BCS, The Chartered Institute for IT.
Focus on and learn from users
By way of an example, he points to one of Deloitte’s biggest charity partners — Scope. The disability equality charity is rolling out PairD, the consulting firm’s generative AI platform.
‘Scope is interested in reducing the digital divide that technology creates’, Sulabh explains. ‘We’re working with them on [among other things] how they can create interfaces that let everybody take advantage of generative AI in an accessible manner. It’s been a real learning experience for us.’
Accessibility enables acceptance and, ultimately, use. To achieve all of this and help team members gain confidence with AI, Sulabh says: ‘The single most important thing is feedback. It’s essential, and the thing is, ease of use is related directly to feedback.’
Summing up, he says: ‘AI is moving from predicting to orchestrating things. AI becomes very valuable if you can reduce the marginal cost of action by automation. This is what’s happening in enterprises — when you automate a process, you’re reducing the cost of action… It’s automation and orchestration of action that drive value.’
