C-Suite Perspectives On AI: Oracle’s Ashok Prakash On Where to Use AI and Where to Rely Only on Humans

As artificial intelligence (AI) continues to advance and integrate into various aspects of business, decision-makers at the highest levels face the complex task of determining where AI can be most effectively utilized and where the human touch remains irreplaceable. This series aims to examine the nuanced decisions made by C-Suite executives regarding the implementation of AI in their organizations. As part of this series, we had the pleasure of interviewing Ashok Prakash. Ashok Prakash is a technology leader at Oracle Health, where he focuses on building AI frameworks designed to improve patient care. With a background in large-scale systems and natural language processing, he began his career as a founding engineer at a logistics startup in India before earning a master’s degree in computer science. His approach to AI emphasizes what he calls “respecting the blast radius,” using automation for low-risk, reversible tasks while keeping humans in charge of high-stakes decisions. Prakash has been an advocate for thoughtful AI integration that enhances human work rather than replacing it, particularly in healthcare and critical infrastructure. He is also the author of Human Reimagined, a book aimed at bridging the gap between AI’s technical realities and public understanding.

Thank you so much for your time! I know that you are a very busy person. Our readers would love to “get to know you” a bit better. Can you tell us a bit about your ‘backstory’ and how you got started?

Thank you for having me. My journey has always been at the intersection of understanding language models and building large-scale systems. I started my career as a founding engineer at a logistics startup in India, where we were solving real-world supply chain problems with technology. This gave me a deep appreciation for how complex systems work in the real world. I then pursued my Master’s in Computer Science with a focus on AI and Natural Language Processing, which led me into core research on how machines can understand and reason with human language.

For the past several years, I’ve been in Seattle, architecting the high-performance cloud infrastructure that powers AI models. My work has now evolved deeper in making AI production ready, and I’m currently at Oracle Health, where we are at the forefront of a new wave of AI. We’re building sophisticated Agentic AI and frameworks designed to fundamentally improve patient care. So, I have this dual perspective: I’m in the engine room seeing how these incredibly powerful tools are built, but my passion has always been about applying them to solve practical, human problems, whether in logistics or now, in healthcare. This journey inspired me to write my book, “Human Reimagined,” as a way to bridge the gap between the technical reality of AI and the public anxiety around it.

It has been said that our mistakes can be our greatest teachers. Can you share a story about the funniest mistake you made when you were first starting? Can you tell us what lesson you learned from that?

This was when I was working at the startup where I was a founding engineer. During the early days, there were nights I used to work very late. One such night, I was working to create a complex invoice template and finally finished it around 3 a.m. I was really proud of my work and, in my tired state, I released it directly to production.

I came back to the office around 10 a.m. to find my co-founder, the CEO, absolutely furious with me. It turned out the invoices generated overnight had a summation error that caused the total value to be in the 6–7 figure range, instead of the 2–3 figures they should have been. From a code perspective, the bug was trivial, I had missed resetting a variable in a loop. But the business impact was huge. It was a humbling and immediate lesson in the immense real-world impact our code can have. But the deeper lesson was about the absolute necessity of human review for high-stakes systems. Automation and speed are powerful, but for something as critical as invoicing, a second set of human eyes is a non-negotiable backstop.

Are you working on any exciting new projects now? How do you think that will help people?

Yes, absolutely. My work has two exciting dimensions right now. First, I’ve focused on building the foundational frameworks and onboarding systems that allow us to integrate new GPU models from different manufacturers much faster. In the world of AI, speed to access the latest hardware is a huge competitive advantage. This framework helps get the most powerful tools into the hands of innovators more quickly.

More importantly, I am now applying this expertise to make AI accessible and usable for healthcare. At Oracle Health, we are building AI-augmented solutions designed to directly impact patient outcomes. For instance, we are developing systems that can help clinicians process information faster and make more informed decisions. We are building an AI co-pilot that can handle data analysis and administrative tasks, ultimately increasing the speed at which we can treat patients and improving the quality of overall healthcare.

Thank you for that. Let’s now shift to the central focus of our discussion. In your experience, what have been the most challenging aspects of integrating AI into your business operations, and how have you balanced these with the need to preserve human-centric roles?

The most challenging aspect is rarely the technology itself; it’s the human element. The biggest challenge is the “AI literacy gap.” There’s a lot of fear and misunderstanding. People either overestimate what AI can do and fear their jobs are obsolete, or they underestimate it and fail to see how it can help them.

The key to balancing this is to shift the conversation from “job replacement” to “task delegation.” We preserve human-centric roles by relentlessly focusing on value. We automate the low-value, repetitive tasks to free up our people for the high-value, uniquely human work. We don’t aim to replace a financial analyst; we aim to eliminate the three hours they spend manually copying data so they can spend that time developing strategic insights. The balance is achieved by making AI a tool that serves human expertise, not the other way around.

Can you share a specific instance where AI initially seemed like the optimal solution but ultimately proved less effective than human intervention? What did this experience teach you about the limitations of AI in your field?

Certainly. In setting up our AI infrastructure, we experimented with using an AI system to speed up our operations by automatically creating rules that could help debug problems faster. The idea was that the AI would analyze system behavior and proactively generate a knowledge base for troubleshooting.

This backfired spectacularly. The AI did exactly what we asked, but without human-level judgment, it created a humongous amount of logs, data, and alerts based on minute, irrelevant correlations. Instead of a clear signal, we got an overwhelming amount of noise. It actually made our lives more difficult by creating a web of data that was impossible for a human engineer to comprehend during a real incident.

We quickly realized this was not the right way to use it. The experience taught me a critical lesson about AI in complex operations: its purpose should be to reduce cognitive load, not add to it. This doesn’t mean AI can’t be useful here, but it needs to be cautiously applied with guardrails in place, designed to surface insights for an engineer, not just dump raw data. The goal is to help engineers navigate the debugging process more effectively, not to drown them in information.

How do you navigate the ethical implications of implementing AI in your company, especially concerning potential job displacement and ensuring ethical AI usage?

This is one of the most important responsibilities for any leader in tech. Regarding job displacement, we are transparent that AI will displace tasks, not necessarily jobs. This framing shifts the focus to upskilling. We have a responsibility to invest in our people, helping them transition from doing the automatable tasks to supervising the AI or focusing on higher-level strategic work.

For ethical usage, we establish very clear guardrails. This includes a strict “data hygiene” policy: no proprietary code or confidential customer data ever goes into a public AI tool. We train engineers to abstract their problems. Furthermore, we mandate that a human must always be in the loop for any high-stakes decision. An AI can suggest, but it cannot decide on things like performance reviews, hiring, or critical system changes. The final judgment and accountability must remain human.

Could you describe a successful instance in your company where AI and human skills were synergistically combined to achieve a result that neither could have accomplished alone?

A great example is in software testing. One of our teams was responsible for a critical legacy service with very low unit test coverage, which made every change risky. Writing hundreds of tests manually was a tedious task that was always de-prioritized.

We implemented a synergistic approach. The human engineer, using their deep understanding of the system, would identify a function to be tested. They would then use an AI assistant to generate the first draft of the unit tests, covering the basic logic and common cases. This took minutes instead of hours. The engineer would then take over, using their expertise to validate the AI’s output, add tests for complex business logic and subtle edge cases the AI missed, and integrate the final suite into our CI pipeline.

Neither could have done it alone effectively. The AI provided speed and scale; the human provided critical thinking and quality control. The result was a massive increase in test coverage in a single quarter, which significantly improved the service’s reliability.

Based on your experience and success, what are the “5 Things To Keep in Mind When Deciding Where to Use AI and Where to Rely Only on Humans, and Why?” How have these 5 things impacted your work or your career?

Absolutely. This is the most important strategic question leaders face today. My perspective is shaped by being in the trenches, from building the underlying platforms to seeing how teams succeed or fail with this technology. Here are the five principles I operate by:

1. Respect the “Blast Radius”: Use AI for Low-Impact, Reversible Tasks. In systems engineering, we always consider the “blast radius” of a change, what’s the worst that can happen if this goes wrong? I apply the same logic to AI. AI is brilliant for tasks with a small blast radius: generating a first draft of a non-critical document, refactoring a function that’s covered by unit tests, or summarizing internal meeting notes. These are reversible and low-risk. But for high blast radius decisions, like deploying a change to a production database, making a final hiring decision, or setting a company-wide technical strategy, you must rely on a human. This principle has become a core part of how I lead teams; it gives us the confidence to experiment aggressively in safe areas while maintaining extreme caution where it matters most.
2. Automate the Toil, Not the Thinking: My work has involved building “paved paths” for hundreds of engineers, and the goal is always to reduce toil, the manual, repetitive work that burns people out. AI is the ultimate tool for this. We should use it to automate boilerplate code generation, dependency updates, and basic report compilation. However, we must draw a hard line at automating thinking. The moment a task requires critical judgment, synthesis of disparate ideas, or long-term strategic thought, it becomes a human-only task. This has fundamentally impacted my work by forcing my teams to focus on what truly drives innovation. We don’t measure success by how much code we write, but by the quality of the systems we design.
3. Use AI for Brainstorming, Humans for the Decisions: AI is a phenomenal brainstorming partner. It can generate ten different ways to design an API, five potential solutions for a bug, or three different architectural patterns for a new service. But it is terrible at understanding the nuanced trade-offs between those options. It doesn’t understand our specific business context, our team’s skill set, or our long-term technical debt. That is a distinctly human, senior-level skill. For my career, this has been a powerful way to mentor engineers. We use AI to generate the “what,” and then we have a deep, human-led discussion about the “why” and “how,” evaluating the trade-offs of each option.
4. For Any System with a Pager, the Human Owns the Final Say: Having been an incident commander for high-severity events, I have a visceral understanding of accountability. My rule is simple: if a system is critical enough that a human engineer might get paged at 3 a.m. if it breaks, then a human must have the final say on any significant change to it. This is especially true in my current work in healthcare, where the stakes are incredibly high. An AI can analyze data and flag an anomaly in a patient’s chart, but the decision on diagnosis and treatment is a profound human responsibility. This principle is a non-negotiable guardrail that allows us to innovate safely.
5. Trust Human Common Sense for the “Why”: My own academic research was in commonsense reasoning for AI, so I know its deepest limitation firsthand: AI is a master of correlation, but it has no grasp of causation. It can tell you what is happening in the data, but it can’t tell you why. For that, you need human common sense and contextual wisdom. I learned this as a founding engineer at a logistics startup, where our algorithm couldn’t understand why a local festival was causing a spike in demand. In any novel situation, a “black swan” event or a bug we’ve never seen before, the intuitive, out-of-the-box reasoning of an experienced human is our most powerful tool.

Looking towards the future, in which areas of your business do you foresee AI making the most significant impact, and conversely, in which areas do you believe a human touch will remain indispensable?

AI will have the most significant impact in areas of scale and speed. It will revolutionize software development by accelerating coding and testing, supercharge data analysis by identifying patterns in massive datasets, and automate most routine operational tasks like system monitoring and repair.

Conversely, the human touch will remain indispensable in areas of strategy, creativity, and connection.

• Strategy: Deciding what problems to solve and setting the long-term vision for a product or company.
• Creativity: True, out-of-the-box innovation and invention.
• Connection: Building relationships with customers, mentoring team members, and leading with empathy. These are the foundational elements of any successful business, and they are profoundly human.

You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people, what would that be? You never know what your idea can trigger. :-)

It would be a movement dedicated to “Democratizing AI Literacy.” The power of AI is incredible, but right now, its full potential is locked behind a wall of technical jargon and complexity, accessible only to a small group of specialists. This is creating a new and dangerous digital divide.

My movement would focus on creating simple, accessible frameworks and educational tools to help everyone, from small business owners and teachers to project managers and artists, understand how to use AI as a tool to solve problems in their own fields. It’s not about teaching everyone to code. It’s about teaching a new way of thinking and collaborating with technology. An empowered, AI-literate populace can unlock an explosion of grassroots innovation that could help solve some of our world’s most pressing challenges.

How can our readers further follow your work online?

The best places to connect with me are on LinkedIn and Instagram. On LinkedIn, I regularly share my thoughts on AI, technology, projects, and career updates. For more accessible life tips aimed at bridging the AI literacy gap, you can follow me on Instagram. You can find me on both platforms at [https://www.linkedin.com/in/ashok-prakash / @thehumanreimagined]

This was very inspiring. Thank you so much for joining us!