KOHb - Getty Images
This is an age of digitization and investment. Enterprises are investing in AI technology to transform their companies.
However, despite efforts to digitize and investments in AI technology, enterprises are still dealing with new challenges nearly three years into the coronavirus pandemic. From changes in consumer buying patterns to employee turnover to supply chain problems, enterprises are seeking ways to deal with these challenges while staying relevant in the age of digital transformation.
Enterprises are not prepared well enough to handle these challenges, said Sateesh Seetharamiah, CEO of Infosys subsidiary EdgeVerve, a vendor of robotic process automation technology. He is also a member of the MIT Auto-ID Lab, an IoT research group.
In this Q&A, Seetharamiah discusses what's keeping enterprises from effectively applying AI technology to their business problems.
What is one of the biggest challenges enterprises face in implementing AI technology on the road to digitization?
Sateesh Seetharamiah: While everybody is dabbling with AI, I think the real deal with AI and all the value it can create is all about having contextual information and data. And if one doesn't have that, the AI's ability to influence cognitive operations and decision-making is going to be limited.
For example, underwriting in the insurance industry. Underwriters need to accurately assess risk, approve or disapprove insurance claims, and approve insurance itself.
Sateesh SeetharamiahCEO, EdgeVerve
The amount of information that they need to gather to make that decision is tremendous: huge monetary, historical information and claims. Many of them are in digital contracts; many have paper contracts and video files. The amount of time that they invest in decision-making is a fraction compared to the amount of time these people invest in collecting the data and organizing the information that they need.
There's a huge amount of opportunity to use technology to digitize information from documents, bring them and build the data layer.
Second, if you have this kind of contractual data, the AI that sits with it can make relatively more intelligent recommendations to these underwriters, in terms of what kind of risk is an acceptable risk. But AI today is hardly used in this process because they don't have relevant data that can be fed into the AI technology to make relevant decisions. So, a very core process of insurance ... is completely manual in human decision-making.
There are many such cases where we see that unless and until everything is digitized, unless and until all that data is very contextual to that enterprise ... it's very hard to really apply AI at a very operational level.
What kinds of processes are needed to apply AI in the digitization process?
Seetharamiah: There are so many policies that govern every single process in an enterprise. AI needs to not only have access to the underlying data, but it should also have access to the key decision-making policies -- and how the policy has been interpreted historically by humans and applied to their own decision-making.
One is the policy, and second is the interpretation of policy and how that has translated to decision-making. That has to come together if AI is to really make recommendations that enterprises can actually use.
Most enterprises don't have the know-how for how processes are executed, let alone automated. Experts in those enterprises have a feel for it, but don't know how these processes are executed across multiple enterprises. Many of them are done even in shadow IT -- it is not even part of the main enterprise landscape. So, that led us down the path of saying we need technology to decipher how processes are executed.
We need to apply technology intelligently to digitize as much as we can. We need to make sure that digital information can be translated so businesses can leverage that at the end of the day. It must serve a larger business cause.
What is keeping enterprises from doing this kind of AI digitization?
Seetharamiah: I think the challenge really lies in having the right kind of governance mechanism between the business and technology, because at the end of the day, that's where real ideas and real implementations and translation of technology to business benefits happen.
It is not just a technology issue. It has a lot to do with internal governance, internal structuring and having the right kind of mindset in approaching this technology. I don't think any enterprise wants to implement even a single dollar worth of technology for the sake of technology at the end of the day. They all have to show a lot of business needs.
Obviously, there is apprehension about some of the stuff AI can leverage, such as, 'Can I hand over this decision-making? Will it affect my job?' There are so many other concerns out there.
There is a larger problem to be solved, which is beyond technology, which has to do with mindset, which has to do with embracing this new-generation technology and convincing people how it not only can influence the business, but can also influence their own individual performance.
Editor's note: This interview has been edited for conciseness and clarity.