As executives navigate digital transformation, integrating Generative AI into strategic planning becomes paramount, necessitating a thorough understanding of its capabilities and a reassessment of existing business frameworks. Generative AI, with its capacity to generate novel content and insights from data, offers avenues for creating value beyond mere productivity enhancements, prompting organizations to explore new strategic avenues. However, leaders must navigate competitive risks and ethical considerations while ensuring responsible adoption and effective implementation. Key focus areas include understanding technology nuances, aligning AI initiatives with business goals, investing in infrastructure and talent, and managing organizational change to foster a culture of adaptability and learning. By embracing Generative AI within their strategic approach, senior leaders can drive significant operational improvements, address complex challenges, and foster sustainable growth for their organizations.
We hosted a roundtable discussion to learn more about this toolkit. The session was moderated by Kashyap Raibagi, Associate Director – Growth at AIM Research and was joined by Hamza Laraichi, Founder and CEO at Infomineo, Mandy Plante, Data and Analytics Leader, Anil Prasad, VP of Software Engineering – Application, Data, AI at Cloudmed, Deepak Jose, Global Head and Senior Director for One Demand Data and Analytics Solutions at Mars, Pankaj Chopra, Vice President for Analytics and Insights at Mondelez International and Prashanth Sarpamale, Chief Executive Officer at Althea.ai.
Enabling Generative AI (Gen AI) at the Senior Business Leadership Level
It’s both the top-down and the bottom-up approach to Gen AI. We operate as a global service company, so it means that our team is composed of people that are native tech-friendly, and whenever there is a new tool or technology, they are eager to test it and evaluate use opportunities. We encourage this proactive tech attitude because we believe it is the most effective way to validate the effectiveness of a new tool or tech.
We ensure that the introduction of Gen AI in our processes is completely reliable first of all from a security standpoint, and with clear guidelines and total transparency for our clients as well. We believe that making it a crowd learning experience will quickly benefit the whole organisation.
At the same time, it’s extremely important that this is enforced and followed through by senior leaders in the organization. In that aspect, we try to make sure that all the key leaders are Gen AI sponsors, and they emphasize its usage following the clear guidelines that we’ve set ourselves in the way we serve our clients.
Data security, integrity, and making sure that we understand the limitations of Gen AI are quite important in the way we work. The overall goal for us is twofold, if you try to simplify it. One is efficiency, trying to make sure that we are able to work in a more efficient way, which helps our team members and analysts free up time to add value to the client and go into a little bit more creativity and innovation in the way they deliver their work. The second element is around quality, making sure that Gen AI enables us to deliver the maximum value added with enhanced productivity.
– Hamza Laraichi, Founder and CEO at Infomineo
The Evolving Role of Data Literacy for Business Leaders in Harnessing Generative AI for Strategic Decision Making
This is something that we continue to have conversations about, almost every day, even outside of the core business. There is so much about the quality of the data, the destruction of the data, what’s available, what’s within the environment – back to the security and other things. All of that is really important.
But then, it’s also about just what the data looks like and what can be made of it, to make it something that’s usable and useful, not just kind of throwing ‘hey, this is the cool shiny object, go use it somewhere’. It’s finding the right problem for it to be the solution for and focusing on that.
The other thing I’ll also add is, in terms of where do we implement it? What’s the right use case, and what we need to worry about? The quality of the data, that’s true – all of those things making a difference. It is a bit of ‘garbage in, garbage out’. If you have a bunch of stuff that makes no sense, isn’t documented, isn’t captured well and characterized well, and then you try to mash it all together, you’re going to get things that are nonsensical.
So there’s a focus there. And the other use case that I actually see on the other side of that is, on top of just having implementing it as part of the enterprise, is actually seeing leaders also use it themselves as almost like a consultant – going and asking questions on what’s the best approach to this type of projects or this particular type of implementation, and using that to then have conversations internally with their leaders and to help with that strategic decision making.
– Mandy Plante, Data and Analytics Leader
Adaptations in Executive Leadership for Embracing Generative AI in Large Enterprises
The concept of democratizing AI is more about fine-tuning existing models. The idea is to take a base model, find your own data, build a customized dataset, and then train your own model tailored to your needs. This is where the current state of democratization stands. The goal is to provide companies the ability to train their own AI models, rather than relying solely on off-the-shelf solutions.
True democratization has not yet been fully achieved. The capability to train your own models is still largely limited to the top players who have the necessary capital, GPUs, and other resources required. For most companies, this level of democratization remains out of reach.
That said, the message seems to be that you can take these open-source models, build your own vector database (which can be a costly endeavor), and then try to derive insights from your enterprise data. But this may not necessarily constitute true democratization, as it still requires purchasing products and adapting them to your specific needs.
The key aspects to focus on are getting the data strategy right, organizing content effectively, upskilling people, and establishing the right framework and architecture to integrate AI into applications. Partnering with cloud providers is also crucial, as seen in recent announcements like Microsoft integrating vector database capabilities directly into their Cosmos DB platform.
However, there is a lack of managed services and support from vendors when issues arise with the AI models built by companies. This self-support model is a concern and a source of confusion for many organizations. Additionally, the lack of clear policies and compliance guidelines is hindering funding and scaling of AI initiatives. Executives need to plan better in this regard, as transparency and monitoring frameworks are still evolving in this space.
– Anil Prasad, VP of Software Engineering – Application, Data, AI at Cloudmed
The Growing Relevance of Responsibility in Generative AI Leadership Discussions
Responsible usage of AI is non-negotiable at Mars. Our Responsible AI principles are closely tied to the company’s five principles. We partner with NGOs like Responsible AI Insititute and tech companies like Microsoft while we craft our AI strategy.
As an AI practitioner, I emphasize the explainability of models. Using a model with inherent biases can lead to significant trouble, often without people realizing it. We are doing specific initiatives to remove bias from the data.
Diversity is a critical aspect. In the Data, Analytics, and AI industry, gender diversity stands at 20%. Even before the advent of generative AI, if we examine the diversity in this group, we find only one woman out of eight in this round table. As part of our AI strategy, we prioritize diversity, and gender diversity is crucial. We’ve partnered with organizations like Women Leaders in Data and AI. Importantly, this commitment starts from the top. Our leadership team (One Demand Data and Analytics Solutions) has ensured 50% gender diversity.
Privacy is another critical consideration. Mars does not collect data from children, and our marketing policy strictly avoids targeting children under 14. These standards are essential for responsible AI practices.
These principles align closely with the core values of our organization. A comprehensive AI strategy must address other aspects as well, including guarding against hallucination or prompt injection attacks, which can significantly impact an organization’s reputation. A thoughtful, comprehensive AI strategy is essential for navigating these challenges.
– Deepak Jose, Global Head and Senior Director for One Demand Data and Analytics Solutions at Mars
Developing GenAI with a Business Lens
We’re talking about Generative AI (GenAI). As an example, from an organization’s standpoint, use of GenAI, may compete with Predictive Analytics, especially when the quality of the data available is inadequate and / or predictive analytics solutions are more appropriate in that context. Hence, it becomes critical to start with the ‘business problem’, which drives the methodology selection versus vice versa.
Sometimes in organizations, there is a race to do something ‘cool’ and ‘trendy’. A few years ago, it was ‘cool’ to build ‘data lakes’ and organizations that jumped into it without a plan ended up building ‘data swamps’: a colossal waste of resources. Fast-forward to today, there is a lot of pressure on analytics teams to work on GenAI use cases. In the absence of a clear business case, its akin to having a hammer in one’s hand and looking for a nail. So as responsible analytics professionals, we need to push back where needed, educate the organizations and do the right thing.
Specifically for the use of LLMs, it will be great for organizations to run small experiments and MVPs and scale up with clear business value has been established. Analytics leaders need to continue to own the solution, work on the right methodologies, as inappropriate methodologies will not provide the expected business value and thereby erode credibility. So do the right thing, not the seemingly ‘cool’ only thing !!
– Pankaj Chopra, Vice President for Analytics and Insights at Mondelez International
Exploring the Slow Adoption of Innovation and Technology in the Healthcare Industry
Healthcare, primarily being a B2B industry , has seen slow adoption of innovation and technology. The focus is more on standardization and interoperability, versus the latest in artificial intelligence. However, conversations on improving both clinical process and operational productivity through deployment of AI have begun – primarily driven by startup disruption
Initial activity has been around productivity improvement – think faster and more accurate summarization, of clinical encounters, of phone conversations – driving quicker authorizations and reviews. These actions reduce costs and improve patient satisfaction. Early successes are leading to more interest around adoption and integration.
Enterprise Data and technology teams are also adopting productivity related AI applications, like copilots to improve the quality of coding and engineer productivity. So the application of AI in the Healthcare sector is very focused on the above productivity related applications – vs purely clinical ones, and rightly so – because clinical applications have a very high bar of accuracy and criticality , in terms of patient safety and outcomes.
Right now, a lot of the AI innovation and disruption is happening in Healthcare technology related startups – the marriage of transactional, claims, and clinical data which sits with large payers and providers, with this innovation brought in by Healthtech startups – is where the magic happens. Early adoption is happening.
– Prashanth Sarpamale, Chief Executive Officer at Althea.ai