Cognizant CEO: “AI will actually create jobs”

The emergence of GenAI has generated both enthusiasm and skepticism. Comparisons are constantly flying around to the appearance of the Internet or the invention of the printing press. Will GenAI live up to these bombastic promises? In conversation with Techzine, Cognizant CEO Ravi Kumar shares his thoughts.

“Any general purpose technology takes time to implement, whether it’s the personal computer, the cloud or the Internet,” Kumar begins. So too with generative AI (GenAI), he assesses. The tech does however have a unique advantage: because it’s “humanizing,” adoption will be faster than other emerging technologies. With this, Kumar refers to the following: whereas an end user would’ve had to delve into the inner workings of a new technology, GenAI lets novel tech talk directly to said end user.

Nevertheless, there are also significant similarities to previous technological breakthroughs. For example, like the Internet or the cloud, AI looks set to continuously improve, lead to innovation in other areas and have a pervasive effect on society and the economy. The latter is something Cognizant has significant sway in. As a global IT service provider, it helps organizations in numerous verticals take their business processes to the next level, from healthcare to manufacturing to banking. It also regularly conducts research; for example, it recently teamed up with Oxford Economics to look at what the future impact of GenAI will look like.

GenAI implementation takes time

GenAI will be adopted quickly because you don’t need any particular skills to use it, Kumar explains. Still, organizations are currently wondering exactly where the business value’s hiding. Even Cognizant itself is still in a prototyping phase, running 250 AI models that will be production-ready at a later date. How do we get to that point? GenAI is creative, but that’s because it is not deterministic: every input generates a potentially unique output through sheer chance. This probabilistic nature makes it difficult to deploy GenAI right now, according to Kumar.

Cognizant is working to eliminate this pain point. It is looking to tweak AI outputs to the point of them retaining their creativity while also remaining accurate. The company is investing heavily in the “explainability” of AI models so that organizations can know concretely how an answer was created. Only with those insights are professionals properly equipped to have GenAI perform business-critical tasks. “Those aspects will mature over time,” Kumar said. “We are taking the raw power of GenAI and making it enterprise-grade.”

Is AI replacing humans?

We often hear that GenAI is going to shake up the job market considerably. A logical concern is that this is simply going to lead to rounds of layoffs. AI is already regarded as the bogeyman that takes away employees’ jobs, even though it is currently mostly about maximizing profitability.

Ravi Kumar emphatically opposes the idea that AI is going to end up replacing humans. He takes issue with the notion of building AI systems specifically to make them a like-for-like replacement for human beings in the workforce. “It’s never going to replace a surgeon,” he says. GenAI can, however, help with diagnostic issues. After all, there’s more data available in all kinds of fields than a human can handle. That applies to both doctors and SecOps engineers, and both benefit from the same GenAI innovations, according to Kumar. It provides a larger knowledge base than the human brain can and recognizes patterns in troves of data that are not manageable for mere mortals.

But how do we get this impact to be quantifiable?

Quantifying AI through an “exposure score” and a “friction score”

Cognizant, together with Oxford Economics, presented the study “New Work, New World.” In it, two metrics appear that clarify the impact of GenAI. A so-called “exposure score” estimates for a given work task how likely it is to be taken over by AI. However, the “friction score” is a promising counterpoint: the higher this score, the more likely it is that this employee can be retrained if necessary.

Kumar cites two extremes: a developer on the one hand and a fisherman on the other. There are 21 million developers worldwide, he notes. AI will shake up their current work tasks, as programming work can largely be done by GenAI solutions, according to Kumar. Since their skills are also a match for a variety of other purposes, they will still be able to secure jobs. Essentially: a high exposure score, a low friction score. With a fisherman, it’s the other way around: good luck finding a robot that completely takes over their work, but what transferable skills does this job contain if at any point AI were to take over?

These scores were applied to 18,000 tasks that keep the economy running, from drafting emails to coding. Oxford Economics and Cognizant looked at 1,000 jobs in which these tasks are done and looked at how GenAI would affect them through 2032. The scores will change over time, because as GenAI matures, the greater the impact on the job market. For organizations, the study can be a prelude to a GenAI roadmap, allowing for timely retraining of staff so they are prepared for an AI-driven future.

Cognizant currently visits companies to assess their exposure and friction scores, among other things. In doing so, it scores all employees. Understanding this, according to Kumar, will become the trigger for organizations to truly embrace AI.

Industries

Cognizant operates in many industries, and GenAI is going to have a diverse effect depending on the industry in question. “Tech for tech,” such as assistance with programming, accounts for about 35 percent of GenAI deployments, according to the company. Content generation and aggregation together account for 45 percent; the other prominent use case at about 20 percent is to assist customer sales reps.

That state of affairs is less awe-inspiring and more utilitarian, but the future may look different. Kumar points to an industry where GenAI is going to give an “immense benefit”: healthcare. “Healthcare is vastly underpenetrated when it comes to technology,” he says. He points out that doctors are so busy with administrative work that, on occasion, they can’t even look at the patient. With its TriZetto platform, Cognizant significantly eases this workload for medical staff. This gives them more time to perform “real healthcare,” i.e. contact with patients.

The deployment of AI will proceed in multiple stages, Kumar said. First it will automate tasks, after which it will eventually transform business processes. However, the right legislation must be in place for this all to happen. Kumar notes that the EU AI Act is a positive course of action, but he stresses the importance of letting innovation roam free. Don’t over-regulate the models, regulate the usage, is the motto here. “If you regulate inputs too strictly, you stifle innovation,” he says.

Need expertise?

The main effect of GenAI, according to Kumar, is that it will shake up how we look at expertise and at various disciplines. Within Cognizant, the top-performing half of the workforce benefits significantly less from GenAI than the remaining half. Whereas the top 50 percent gain 15 percent in performance, the figures for the other 50 percent jump up by 35 percent. In short: GenAI can bring the performance envelope closer together. This will democratize knowledge and increase social mobility significantly, is the Cognizant CEO’s firm belief.

It also bridges disciplines. For example, the best financial analysts are said to have a Liberal Arts background, not one in mathematics. But how do you get those people to take on that job if they have no knowledge of statistics? That’s something GenAI is going to be a supplement for, Kumar says. “New programmers don’t have to be able to program”, either. It’s an optimistic view of GenAI’s development, which, incidentally, he shares with Nvidia CEO Jensen Huang. The latter even advises developers against learning to code today. That’s a rather extreme viewpoint. Won’t we always need specific expertise? After all, who will verify that the AI generated code makes sense, or that an AI chatbot correctly describes academic sources?

“The need for expertise will not go away,” Kumar acknowledges. But GenAI also builds expertise precisely where none was previously available. For example, Cognizant is working on a platform called BioNeMo from Nvidia, which facilitates building AI models for drug discovery. Such a model will eventually be able to understand and describe DNA better than a human. Biologists can talk in their own language to a chatbot that in turn “speaks” DNA natively, as it was trained on the data that reveals the workings and “grammar” of DNA.

AI, according to Kumar, is additionally going to emerge as a connecting element between professions that currently operate separately. This makes it “cross-functional”: a software engineer and a medical specialist can communicate better with each other through AI by having their own professional vocabularies converted to language that the other can better grasp.

Conclusion

Ravi Kumar is a self-proclaimed tech optimist: he sees extraordinary opportunities for GenAI. Despite challenges such as bias and regulations, he clearly believes in an clearly positive impact of GenAI on the labour market and society. His company contributes to that mission as well, in addition to the fact that Cognizant encourages AI adoption through its own research and contact with clients. Ultimately, AI should provide quantifiable improvements. It can only deliver on that promise with hard numbers and continuous innovation, and that’s something Cognizant wants to contribute to.

Also read: Bill Gates sees no problem in growth of energy consumption due to rise of AI