30 Under 30 2024

'As Gen AI goes enterprise mainstream, expect a "value-first" shift in IT services': TCS

Harrick Vin, chief technology officer, and Sivaraman Ganesan, head of the AI.Cloud business unit at Tata Consultancy Services, talk about the company's AI and generative AI strategies

Harichandan Arakali
Published: Jan 24, 2024 02:53:59 PM IST
Updated: Jan 24, 2024 03:09:50 PM IST

'As Gen AI goes enterprise mainstream, expect a "value-first" shift in IT services': TCSHarick Vin(sitting), CTO, and Sivaraman Ganesan, head of the AI.Cloud business unit at Tata Consultancy Services Image: Bajirao Pawar for Forbes India

With software and artificial intelligence (AI)-based automation set to accelerate through this year and beyond, the aim is to augment customers' ability to get more from technology in every aspect of their operations and help them become owners of creativity, Harrick Vin (HV) and Sivaraman Ganesan (SG) tell Forbes India in a recent interview. Edited excerpts:

Q. Can you give us a snapshot of how work on AI has evolved over the years at TCS?
HV:
The field of AI has witnessed extensive development over several decades, encompassing three primary intelligences. Recognition intelligence involves extracting meaning from unstructured data like images or text, a domain in which we've been active for at least 15 years, focusing on tasks like sentiment extraction and image analysis for damage assessment.

Reasoning intelligence, the second type, combines descriptive, diagnostic, predictive, and prescriptive aspects. This area encompasses traditional machine learning models and analytics, as seen in products like Ignio, designed for AI Ops.

Lastly, generative AI represents the third intelligence, focusing on operationalising insights and prescriptions derived from recognition and reasoning. It entails converting information into human-readable reports or automating procedures to address issues, marking a natural progression toward closed-loop intelligence.

This evolution aligns with the goal of translating AI-derived insights into actionable outcomes.

Q. From a customer's point of view, what’s changed?
 SG:
In the span of two decades, particularly in the early days, the focus revolved around organising data efficiently and extracting intelligence from the structured information. This journey involved creating databases, refining data schemas, and utilising advancing computing capabilities, especially through big data techniques.

The market demand dictated the creation of structures to assist businesses in querying for intelligence and deriving insights or analytics. Driven by the evolving landscape, the current phase introduces generative AI, a novel approach that not only encompasses traditional tasks but also integrates co-pilots and expedites decision-making processes. Validation through human involvement remains essential.

The discussion also delves into the impact of newer chipsets, GPU technology, and heightened computational power, significantly enhancing the speed at which enterprises can leverage these technologies for mutual benefit. This ongoing evolution represents a notable advancement in the synergy of data organisation and intelligent decision-making capabilities.

HV: Engaging with emerging technologies like large language models and Gen AI has provided valuable insights. A notable implication is the potential dominance of natural language as the universal interface for machine interaction, possibly rendering natural language as the sole interface needed.

However, we've discovered the necessity to blend predictive AI with generative AI to unlock substantial value. Recognising this as an augmentation game rather than a replacement, we understand the technology's real power lies in propelling individuals, teams, and organisations from good to great.

The journey is complex, demanding meticulous preparation in terms of data, environment, and the creation of purposive agents tailored to specific tasks or activities. Choosing the right mix of intelligences, such as large language models or predictive AI, involves numerous decisions, making the solution-building process intricate.

Beyond being a technological challenge, it unfolds as a change management issue, fundamentally transforming roles within organisations. As co-pilots enter the scene, individuals transition from doers to trainers of machines, handlers of exceptions, and owners of creativity.

This dual nature of technological and organisational complexity underscores the intricate path toward truly deriving value from these transformative technologies.

Also read: TCS in 2023: Five headlines for the rear-view mirror


Q. Give us a sense of how the combination of your human professionals and your AI capabilities delivers this augmentation to your customers.
HV:
Our strategic approach is clear: harness existing resources from open source or partners and build advanced layers of intelligence to expedite solution development. We've conceptualised this strategy across four distinct layers, collectively termed "IT for AI."

The first involves facilitating organisations in their transition to AI-first enterprises, focusing on accelerating the AI journey. The second layer, "AI for IT," centres around the transformation of IT, DevSecOps (A software development practice that aims to build security features early on in the development process), and similar domains through the infusion of AI.

Moving forward, the third layer, "AI for AI," addresses the complex challenge of constructing purposeful agents using a combination of predictive and generative AI, emphasising efficiency, effectiveness, and responsibility.

Lastly, "AI for business" explores how AI can redefine both horizontal and vertical business functions, adapting to industry-specific or common functional requirements. Our focus revolves around combining partner and open-source contributions with additional layers of intelligence to simplify and drive innovative practices in these four dimensions.

SV: The TCS services strategy is actively materialising, with recent quarters witnessing the implementation of various team elements. Engaging in assist and augment use cases, or what we term purposive agents, has become prominent. These atomic use cases, focused on efficiency and workflow enhancements, have garnered substantial interest, adoption, and dedicated efforts.

Concurrently, discussions have shifted to how AI aids software engineering, delving into potential tweaks within the software development lifecycle (SDLC). Yet, the broader narrative extends beyond technological nuances to encompass elevated conversations about business possibilities.

Customer board discussions are emerging, signalling a shift towards addressing potential transformations enabled by AI. While not yet mainstream, these dialogues indicate a trajectory where the needle moves from basic applications to more intricate possibilities. Acknowledging the infancy of this journey, the evolving landscape promises significant potential for our ongoing efforts.

Q. Tell us more about this. What are your customers asking?
SG:
To summarise in a few bullet points, what customers are asking for, one is using Gen AI and to some extent using data on the cloud, what is the art of the possible? What are those use cases that we can get going quickly to illustrate how we can consume this technology?

The second ask is at the other end of the spectrum. Using AI, not necessarily Gen AI alone, and everything that the cloud has to offer, how do you reimagine or transform our business? From a bigger value chain perspective, what's the art of the possible? And then the third is you have services going, you have solutions being done for us over the years, over decades.

How does AI now come into the picture and what role does it play for you to do that in a much more optimal, much more efficient, much more quicker manner?

Q. At a high level, do you see accelerated automation of IT and a bigger shift towards delivering business value from tech?
HV:
The nature of work is poised for a redefinition, not replacement. Consider software DevOps: the current emphasis on script writing and infrastructure provisioning will diminish as automation takes over. Engineers will shift left, directing their focus towards product requirements analysis, acceptance criteria definition, and adaptable software and architectural design. (Shift left refers to moving the testing, quality, and performance checks to much earlier in the software development process, often even before any code is written.)

This shift emphasises critical thinking, strategic planning, goal setting, and creative problem-solving skills, aligning with the concept of being owners of creativity. The transformation in IT roles doesn't imply automation; instead, it grants individuals time for higher-value tasks, augmenting their end-to-end capabilities.

Moreover, this revolution prompts a shift towards a value-first mindset, where business problems are decomposed into smaller challenges, each requiring specific intelligences and augmentations. This approach aligns with the emerging trend of business value-led technology innovations, emphasising the importance of aligning technology solutions with business needs.

Overall, the evolving landscape requires continual adaptation, with the role of individuals evolving in tandem with technological advancements.

SG: The whole shift-left agenda is indeed, I think, where it's all headed. Historically, if you take the example of coding, it was all manual. And then came along IDEs (integrated development environments). Then came along an element of automation and instrumentation.

And now through prompt engineering, there's a new possibility. Now, does that mean coders in this example get substituted? I would argue they get supplemented. And I think that is principally the mantra here. How do we get what is deemed basic stuff, or if I may, happy path stuff, how can that be done speedier, quicker, through Gen Ai and techniques like that?

 And how do you look at outliers, exceptions, or that zone of creativity and augmentation where a human pretty much still needs to be in the loop, if not needs to be the driver. And if you apply this philosophy to every node of the value chain leading to business problem formulation and business solution definition, to therefore the tech design and everything else that ensues, then I think you have what would be called a supplementary effect as opposed to a substitutional effect. And I think that's where we are at this point in time.

Post Your Comment
Required
Required, will not be published
All comments are moderated