Q&A: Accenture’s Prasad Talks AI Adoption for the Enterprise

Where artificial intelligence fits into an organization’s strategy can be a challenge to nail down. The technology can be applied in a host of ways that might speed up certain tasks and processes, but only if leadership has a firm grasp of what is needed and how it can be applied. Rajendra Prasad, senior managing director, global automation lead for Accenture Technology, was in town this spring for the O’Reilly Artificial Intelligence Conference in New York where he provided some perspective on the matter.

He took time out with InformationWeek to discuss how enterprises should approach introducing AI and then scaling up their application of it.

What are some of the opportunities that AI offers to enterprises?

“AI is no longer [just] a buzzword. It is now real. A lot of people mix AI, automation, and machine learning together. I have a different view of this. Every large enterprise goes through a journey of evolution and automation. The most important thing in that evolution is to understand why you’re doing this. Don’t do an AI if it is not required. If you can fix a problem without AI, just do it. That’s the first thing. The second thing is, I learned a long time back, if you don’t know where you are on a map, a map won’t help you. It is very important for enterprises to know where they are in the AI journey map before they put forth a plan.

Image: metamorworks - Adobe Stock

“This is nothing new; you do this for everything. When a new technology like AI pops up, people believe they should get on board with it without thinking. That’s the first thing enterprises have to deal with. They have to think structurally about where they are on the map, how to get to where they want to get to, and then follow the principles of software engineering. We developed within our organization a very structured assessment and appraisal technique to quantitatively and qualitatively evaluate enterprises regarding AI. Then you put together a plan: You say you want to do deep learning, machine learning, neural networks, and what use cases you want to do. You have to put together a proper plan, like you do for any change program.

“This is what I call simple thinking. The idea is to simplify everything that you want to do.”

Where does the conversation about simple thinking need to happen within an organization? Who are the stakeholders who should be involved?

“Typically, in large enterprises the change is driven at the top. In this case, the top management can be the CEO, the CFO, procurement chief -- whoever can decide what needs to be done. Even if you get that commitment, the most critical part of this ‘why?’ What is the business relevance?

“When we implement AI in our company, people have to ‘deposit’ the value they are going to generate by using the AI solution in their enterprise. That is treated as a debit. As you progress through, you need to show the business benefits of implementing that and then you take out the ‘money’ you put in. When you cross the deposit threshold, you start providing an ROI where you are getting the benefit of implementing an AI solution.

“People ask, who should be the sponsor of this? Is IT? Is it the CEO? My answer is, anyone who can influence the organization. Anyone on the leadership team who can impact the behavior of the organization. If you don’t find such a person to be a sponsor, the other way to do it is simple. In any organization, there is an entity that is going to make money for you. There will be someone within that moneymaking entity who can have an impact. They should be the champions for change.

What are some pain points within organizations and the corresponding types of AI that can provide solutions?

“In today’s world, most enterprises try AI based on use cases. The way they generate use cases is based on their business needs at that point in time. That is like a chicken-and-egg situation. The most important thing is how to automate that use case in a relevant way for business.

“Whenever you build an AI application, at the end of the day an AI application is a piece of software. All the rules of software engineering apply. What are the rules of software engineering? There should be methodology, there should be a process, there should be a procedure, there should be security considerations. If you follow those principals, you will automatically address some of the challenges AI has today. There is one dimension of the challenge that is specific to AI, which is the data. AI is the user interface for data. The data drives AI.

Where are organizations looking, or where can they look, to build up their use of AI? In-house? Externally?

“At the end of the day, the success and sustainability of AI capability within the enterprise will only come if the organization builds its own capability. We built our own automation and AI career model. We have curriculum that everyone in the organization has to go through and graduate to new levels. If you don’t do that, we would not get the talent. There is a limited pool of talent. It’s a supply-demand issue. You can start by depending on external resources for a while, but like any change program success will only come if you build internally.”

For more about AI in the enterprise check out these recent articles:

Automation, AI Impact: The Future of Work

ERP Giants SAP and Oracle Add AI to Platforms

Will the Real AI Please Stand Up?

9 Steps Toward Ethical AI