The secret to successful AI outcomes? It’s data, says Hitachi Vantara CTO Jason Hardy

Hardy sat down with indianexpress.com to share insights into the shift from AI to physical AI, the approach of leaders towards data and AI governance, and challenges and opportunities for organisations. 

What is ‘good enough’ data for enterprises?

“That’s actually a really funny question. What is good enough? Because good enough is really up to each person,” he said. “Good enough could be 75 per cent accurate, or good enough needs to be 95 per cent accurate, or 100 per cent.” 

The executive shared that data context matters enormously. He contrasted by sharing that, for instance, if AI is added to a power grid to manage the power system, it would need to be 100 per cent accurate. Conversely, when it comes to customer service applications, 80 per cent accuracy may suffice. This shows that AI deployment is not one-size-fits-all but rather a careful evaluation of use case requirements and available data quality.

Moving from experimentation to outcomes

The AI landscape is, at present, undergoing a dramatic shift where companies are looking towards outcomes now more than ever. Hardy noted that the industry has moved away from endless proofs-of-concept, partly driven by financial pressure. “A tremendous amount of money has been spent on this, so we need to see the ROI (return on investment),” he observed. “We are now being forced out of this forever experimentation and now into this thing that needs to start being able to pay for itself.”

However, Hardy also offered a contrarian view on the conventional ROI framework. “Because AI is driving so much transformation in business, one could argue that the failures are actually more important than the successes, because it is helping you understand the weak points of your business and how you need to transform from a digital perspective,” he said. This point of view reframes failure not as wasted investment, but as valuable organisational learning.

At the same time, a lot of companies are introducing pilot projects, with many being stuck in the demo stages and many unable to move into production. When asked what could be the single biggest challenge for enterprises, Hardy said, “The biggest challenge here is that there are no clear goals that were defined when you started the pilot or the proof-of-concept.”

He explained that organisations often launch initiatives without properly evaluating whether success is achievable, considering their data maturity, infrastructure, or technical capabilities. 

The move towards physical AI

When asked how he would describe physical AI, Hardy said he calls the shift ‘conveyor belt to cloud.’ He simplified it, saying, “Where we can take AI’s capabilities and have it represented in the physical space.”

While traditional manufacturing robots are programmed for single repetitive tasks, physical AI systems are capable of demonstrating environmental awareness and complex problem-solving. “It’s now literally an AI brain in a physical form that actually manipulates space around it,” he noted, alluding to humanoid robots and autonomous vehicles such as Waymo and Tesla as early examples of this transformation. 

When asked if there is an AI momentum in India, Hardly admitted that he sees India as particularly well-positioned for AI adoption. “Indian society has been digital first for a very long time, so naturally they are more ready to adopt something like this at a societal level,” he observed. The country’s digital infrastructure for payments, passports, and identification systems has created a technologically savvy population.

Moreover, government investment in sovereign AI initiatives is democratising access to technology. “Traditionally, this technology is very much for the largest of companies. Now that’s being torn down through these sovereign AI initiatives,” Hardy said, adding that it enabled startups, students, and smaller companies to innovate without prohibitive infrastructure costs.

The age of AI and leadership

When asked what his advice would be for a Chief Information Officer (CIO) in India starting their AI journey today and what he would tell them not to do, Hardy offered practical yet philosophical insight. “AI is not a product. It is an outcome. It is a way to answer a question.” Secondly, leaders must assemble diverse, cross-functional teams and give them autonomy. “You need to put them in a room and leave them alone. Let them think it through, collaborate, figure out the ideas, and innovate.”

He also warned against impatience. “It’s like planting a tree. You are not going to get an orange overnight. You need to let it take root and then shoot up.” On the other hand, for those beginning their AI journey, Hardy offered clear counsel: target the “low-hanging fruit”, or problems that are easiest to solve but deliver high value. “Don’t try to transform your business from day one,” he cautioned.

The Hitachi Vantara executive’s unique perspective cuts through the AI hype cycle with pragmatism that is rooted in data science fundamentals. His message is clear – successful AI adoption requires quality data infrastructure, realistic goal-setting, patient leadership, and an insatiable willingness to learn from both successes and failures.