Transforming Role of AI in IT Development and Implementation

Recently siliconindia interacted with Aanand Krishnan and Vijay Kumar from Oracle to discuss about AI technologies which have stormed the technology world over the last few years. It is changing the way people are building their applications. But, how they're going about the process of building, the process of deploying, process of maintaining applications is fundamentally changing because of AI. Aanand Krishnan, Vice President of Product, Oracle Cloud has two decades of Silicon Valley experience in building world-class products and providing strategic advice to executive management on a wide range of topics - product strategy, long-term planning, M&A and corporate strategy. Vijay Kumar, Vice President - Product Marketing, Oracle is managing product marketing for Application development services and Developer Relations.

What are the primary concerns for the developer in AI code development?

There are four different things that are the primary concerns of the developer:

Writing code: Writing repeatable pieces of code consistently time and again in a way that it's error free. It’s also performance optimized for mundane tasks like writing it over and over again.

Debugging: It is an onerous task. If you write a code, you have to go find if there are bugs. If there are codes written by others that you want to improve on, there may be bugs that you want to find.

Documentation: Millions of lines of code written over the years or new code that they're writing. You want to write good documentation so that others can read it. Others can learn about the code and the application of the product in general. 

Testing: Whether it's regressive testing, unit testing, feature testing or similar subjects.

What is AI integration and what does it mean for the developers?

It is basically a tool, a plugin that exists in a development environment. As developers are using a development environment to write code, to build applications, this tool runs inside that development environment. But it is powered by the large language models in the cloud. So, in that sense, it is AI powered. If the developer wants to build new code, he can either ask the assistant to build a code snippet, copy it into his application, or he can ask a chat-based question about coding, about code, or about anything related to coding, programming. Those are the kinds of things it does for developers. So, it makes their job easier as they're building applications.

What are the barriers that are coming in the way of implementing AI in industries?

AI is still a new technology and enterprises are figuring out exactly how to get value out of AI or generative AI. So, we're still in the first innings of this journey. But we are seeing a number of use cases in which generative AI is playing a significant role and has a lot of potential. One is in the area of human capital management, particularly recruiting, where generative AI can play a very significant role from soup to nuts. Gen AI is able to write really attractive job descriptions, and even do screening in an automated way, pass the candidate along so that one can automate and screen and schedule candidates much faster. Another area where we see generative AI play a very significant role is in customer service, chatbots and digital assistants powered by generative AI using things like retrieval, augmented generation are actually becoming quite successful. So, I think there are lots of use cases across the board, but early indications are that customer service and human capital management,  recruiting are areas where you're going to see a quick ROI.

What are the barriers faced by customers in implementation of AI into their industry? 

The pain points that the customers are seeing in implementing AI are data quality, data quantity and corpus of data to implement. That's one of the areas where we are well positioned to address the issues. The second thing is the lack of skill set to build AI applications and things like that.  Where a lot of this code assistance and a lot of other expertise is needed that technology companies like Oracle can bring. Third thing, of course, is obviously, a top of concern for most enterprises is cost that includes infrastructure, the GPUs and everything else. It's one of the top concerns that we often see customers face. 

But bringing it back to what we are talking about here, I think this specific product is intended to address the developers' ability to build applications faster. Human capital is one of the hardest ones, especially skilled human resources are hard to find.

What are the future advancements in AI and machine learning technologies? 

It is hard to predict how this market is going to perform. We can see that the pace of innovation in AI is faster than ever before. If you look at the kind of text generation, image generation, video generation capabilities we had 12 months ago compared to where we are now, it looks like you're watching an old Charlie Chaplin movie compared to something like Mission Impossible. It's like night and day. So AI is evolving at a fast pace. It would be very difficult to predict how this is going to play out, especially because, when generative AI first came out, even the researchers didn't quite understand exactly why it did so well. So I think we are in a very interesting space and in a very interesting time where technology is evolving at an incredible pace. There will be a lot of innovation. Enterprises are now starting to realize the importance of AI. But I think prediction is a tough game.  The best thing we can do in this space is keep pace, iterate fast and listen to customers closely.

Will AI applications be able to replace human brain applications in industries?

It’s not like one day will come when AI will suppress human brains. When one is debugging any software, coding or performing any other relevant task, brain usage is primary. But if the AI is doing all those things, we are not doing that. The demand for software will only grow and new types of roles are going to get created. Overall, if we think about the lifecycle of a developer, at the end of the day, you can't replace all of it. You still need a human in the loop. So, the nature of development will change. The demand for software and the velocity with which software gets released will change. But any notions that software will completely replace developers are a little bit premature.