Rohit Pandharkar, a partner at EY is a distinguished leader in data science, SaaS, and artificial intelligence (AI). With a notable career spanning at EY India, Circles.Life (Singapore), OLX Autos, and Mahindra Group. Specializing in Generative AI, he has extensive experience in AI, computational imaging, and cryptography. He is recognized for his contributions to international journals and conferences and has been honoured with prestigious awards like Data Science Professional of the Year. 

Rohit advocates for responsible AI adoption, emphasizing cybersecurity, privacy, and fairness to enhance human experiences and drive societal well-being. 

In a career completely devoted to Data Science, SaaS, and the Technology sector, your journey has been unique? 

It’s been a unique journey and my career has been dedicated to the intersection of Data Science, SaaS, and the Technology sector. Firstly, it’s crucial to understand one’s skills, strengths, and interests. This involves knowing what you can do, what you enjoy doing and seizing the right opportunities. The ideal scenario is when these aspects align perfectly—a concept known as Ikigai. For me, this alignment is evident in my passion for AI and data science, where I’ve cultivated expertise and now advise and execute AI transformations across diverse sectors. 

Secondly, it’s essential to anticipate industry trends early on. I recognised the emerging wave of deep learning and AI in 2015, shaping my career path to become a leader in data and AI for large organizations. This foresight has led me to pivotal roles at Mahindra Group, OLX Autos, and now at EY, where I drive data science and AI initiatives. 

Lastly, developing a ‘T-shaped profile’ has been instrumental—gaining broad exposure across sectors like finance, manufacturing, e-commerce, and telecom while specializing deeply in data science and AI. This breadth and depth have equipped me to apply AI technologies effectively in various business contexts. 

What have been some of the challenges and lessons that you have learned in your journey? 

The field is evolving rapidly, presenting challenges that require continuous learning and practical implementation to maintain credibility among industry leaders. Staying updated involves delving into the theory and algorithms behind generative AI, experimenting with open-source tools, and hands-on coding to run proof-of-concepts. This demands significant time and effort to substantiate expertise beyond mere theoretical knowledge. 

Another challenge is the multifaceted role of an AI leader. Apart from technical proficiency, one must excel in talent acquisition, development, and project execution while effectively

communicating AI strategies to business leaders, CXOs, and boards. Articulating how complex models translate into business impact and delivering on these promises is intricate and time-consuming. It’s a balancing act between nurturing talent, designing projects, and conveying the vision to stakeholders while demonstrating tangible ROI. 

Successfully navigating these responsibilities is no easy feat, requiring a blend of technical prowess, strategic communication, and managerial acumen to drive meaningful AI initiatives. 

Considering your deep knowledge of artificial intelligence and in particular of generative AI, can you tell us about the developments in that field over the past few years? 

AI and deep learning have gained significant traction in the past decade due to advancements in computing speed, affordable GPUs, and refined algorithms. The digitization of data through cloud and mobile technologies has further fueled innovation. Notably, ImageNet competitions demonstrated AI’s ability to surpass human accuracy, paving the way for transformative technologies like driverless cars. 

Post-2017, Google’s “Attention is All You Need” paper on transformer architectures marked a pivotal shift in AI capabilities. Transformer models like ChatGPT revolutionized conversational AI, enabling seamless interactions in multiple languages. ChatGPT’s public release in November 2020 quickly gained over 100 million users daily. 

Generative AI goes beyond language, creating content across media formats—images, text, sound, video, and code. Tasks that once required deep expertise, like generating Fibonacci series code, are now accessible through simple prompts. 

Generative AI’s impact extends to business applications, facilitating customer interactions and optimizing processes in sectors like automotive risk management, finance customer engagement, or FMCG analytics. 

In summary, generative AI represents a paradigm shift, enabling intuitive interactions and driving business value across diverse industries. 

What have the adtech and other industries done to adapt to this technology? 

Ad tech and other industries have embraced generative AI to streamline and enhance various processes. In ad tech, leveraging generative AI means having an intelligent tool capable of generating and optimizing content based on simple English instructions. 

For instance, creating ad banners is now accessible to anyone with the ability to describe their desired ad in English. This eliminates the need for specialized software or artistic expertise previously required for image or artwork creation. The approach has evolved into “prompt engineering,” where clear and precise instructions are given to the AI to ensure relevant and

high-quality output while safeguarding data privacy and integrity. 

Similarly, the capabilities of generative AI extend to video and music production. Startups like Mubert and Beatoven.ai enable the generation of music compositions and background scores with English instructions, simplifying creative processes across industries. 

Generative AI also acts as an orchestrator for multiple systems, including ad tech platforms. Advertisers can achieve targeted outcomes efficiently by instructing the AI to optimize ad spends across various channels while adhering to specific constraints (e.g., daily budget limits, channel composition). This includes ensuring ad relevance and compliance with spending guidelines across different platforms. 

The advancement of compound AI systems further enhances these capabilities, allowing for sophisticated testing and optimization strategies within ad tech and beyond. In summary, generative AI has transformed content creation and optimization processes across ad tech and other sectors, democratizing creativity and efficiency while addressing complex operational challenges. 

In your experience working in the APAC region, what similarities and differences have you observed? What technological advances have they adapted differently or similarly? 

In my experience working across the APAC region, I’ve observed varying levels of digital maturity and sector dominance among different countries. Singapore, for instance, demonstrates significant tech maturity and population savviness, particularly in financial services. India also exhibits high digital savviness, leveraging platforms like WhatsApp, Instagram, and UPI extensively. However, sector prominence differs; financial services dominate in Singapore, while India’s digital adoption extends broadly. 

One key difference lies in the potential business impact of technological advancements. For example, in India, initiatives like Aadhaar (digital ID) combined with AI-enabled services such as airport face recognition (e.g., DigiYatra) significantly enhance efficiency and productivity. This innovation saves millions of people valuable time and boosts overall happiness and productivity, vital for a nation’s well-being. 

India’s digital public infrastructure, including the JAM stack (Jan Dhan-Aadhaar-Mobile), UPI, and ONDC (Open Network for Digital Commerce), sets it apart in the APAC region. This integrated framework, serving over 1.4 billion digitally-savvy citizens with affordable mobile data, fosters economic growth and enhances quality of life. 

In contrast, readiness among citizens and government infrastructure varies across countries like Cambodia and Indonesia, impacting technology adoption and innovation. India’s unique digital ecosystem exemplifies transformative possibilities, benefiting businesses and enhancing societal well-being on a vast scale.

With LLM models and artificial intelligence chatbots becoming smarter and better, do you believe that these advanced LLM models will soon be able to understand semantics? LLM models and artificial intelligence chatbots are indeed becoming smarter, raising questions about their ability to understand semantics. From a technological perspective, this concept is often evaluated through the Turing test, established by British computer scientist Alan Turing in 1950. The test assesses whether a machine can engage in conversation indistinguishably from a human. 

Today, with advancements like ChatGPT, Gemini, and Microsoft Copilot, some argue that the Turing test has been passed, while others remain skeptical. The ability of AI models to converse seamlessly with humans is approaching a very human-like level, blurring the line between AI and human interactions. 

Notably, companies like hume.ai have developed empathy bots capable of adapting their voice and responses based on the user’s emotions. These bots demonstrate conversational emotional intelligence by adjusting tone, diction, and word choice to match the user’s mood. 

However, achieving artificial general intelligence (AGI), where AI can perform tasks comparable to humans across diverse domains, remains elusive. Predictions about AGI range from skepticism to anticipation within the next several years, highlighting ongoing debates about its feasibility and potential implications for humanity. 

Ultimately, the progression of AI towards understanding semantics and human-like interactions represents a significant milestone in technological development, sparking both excitement and contemplation about the future of AI. 

While the Indian advertising industry is gradually adapting to better DOOH opportunities, it lacks creativity and seamless technological integration. Is there a chance that the Indian market will soon outperform other international markets on this front? 

When considering creativity and technological integration in the Indian advertising industry, it’s important not to directly compare it with other international markets. Each country’s advertising landscape is unique, shaped by factors such as economic size, dominant sectors, and the maturity of marketing technologies. 

India, however, stands at an intriguing juncture. With a large, trainable, and tech-savvy population—leveraging technologies like AI bots for image and video creation—we could witness a surge in creative output. This potential adoption of AI-driven creativity by businesses and brands could lead to rapid integration, driven by its significant impact on business ROI. 

The demographic dividend in India, coupled with a highly trainable population embracing Gen AI

technologies, opens up vast possibilities in marketing and advertising. As this skill base expands, India’s advertising industry could see transformative growth, leveraging creativity and technology to drive impactful marketing strategies. 

While technological advancements are revolutionizing our lives, there is also the threat of carbon emissions that persist. How can carbon emissions be controlled without hampering progress? 

Controlling carbon emissions with advancing technology is a challenge. Data centers that power AI models use a lot of energy, which can contribute to pollution. However, companies are making strides in sustainability. They’re improving data center efficiency and using renewable energy sources like solar and hydrogen. 

Another area of progress is in chip design. New chips are more energy-efficient, reducing overall energy use. Additionally, advancements in machine learning algorithms, like tiny LLMs (Language Model Models), enable effective model training with fewer parameters and data iterations. Smaller models can achieve good performance with less data and fewer computations. 

Combining these efforts—like using cleaner energy, better chips, and smarter algorithms—can lower the environmental impact of AI technology. This balanced approach is essential for sustainable progress without sacrificing the benefits that AI brings. 

Will the hype surrounding AI continue or decrease in the future, considering its prevalent use over the last five years? 

The term “AI” encompasses various areas like robotics, AI chips, algorithms, and applications. Within AI, there’s machine learning, which focuses on software algorithms to predict patterns, and deep learning, which mimics human brain neurons for predictions. Generative AI, a subfield, allows AI to create images using transformer architectures. 

As for the hype around AI, technologies typically undergo phases like nascency, hype, and then either obsolescence or mature adoption. AI is likely to become ingrained in our lives but needs responsible implementation. Guardrails for cybersecurity, privacy, fairness, transparency, and accountability are crucial for mature AI integration. Ultimately, AI should enhance human experiences and happiness, not dehumanize us.