Three implementation options for Generative AI: An interview with Pedro Uria Recio, chief data analytics and AI officer at True Digital
In this episode of the Heidrick & Struggles’ Leadership Podcast, Heidrick & Struggles’ Matthias Kamber speaks to Pedro Uria Recio, the chief data analytics and AI officer at True Digital, the digital arm of Thailand’s leading telecom company, and managing director of True Analytics, a corporate venture that builds data-driven enterprise solutions in advertising, credit risk, customer intelligence, and data enrichment. Recio shares what drew him into the AI space and talks about key business applications for generative AI, as well as what new skill sets and capabilities organizations will need to seek in leaders to maximize opportunities and minimize the risks. He also shares his thoughts on the regulatory, legal, and ethical questions AI is bringing to the forefront of society and how he believes the job market will change in response to these technologies.
Some key questions answered in this podcast include:
- (1:31) Can you talk us through your journey into technology and what drew you into the AI space?
- (2:41) There is a lot of hype about generative AI since ChatGPT was opened to the public late last year. Can you tell us what generative AI is and talk about key business applications for it?
- (10:30) What new skill sets and capabilities will organizations need to seek in leaders to maximize their opportunities?
- (16:37) Generative AI raises several regulatory, legal, and ethical questions for businesses and society. For example, how can ethical guidelines be established to ensure accountability and transparency in developing and using these systems?
- (20:20) How do you see the job market changing due to generative AI? (20:20)
Below is a full transcript of the episode, which has been edited for clarity.
Welcome to the Heidrick & Struggles Leadership Podcast. Heidrick is the premier global provider of senior-level executive search and leadership consulting services. Diversity and inclusion, leading through tumultuous times, and building thriving teams and organizations are among the core issues we talk with leaders about every day, including in our podcasts. Thank you for joining the conversation.
Matthias Kamber: I’m Matthias Kamber, a principal in Heidrick & Struggles’ Zurich office and a member of the Global Industrial Technology Practice. In today’s podcast, I’m excited to be joined by Pedro Uria Recio, chief data analytics and AI officer at True Digital, the digital arm of Thailand’s leading telecom company, and managing director of True Analytics, a corporate venture that builds data-driven enterprise solutions in advertising, credit risk, customer intelligence, and data enrichment. Pedro manages large-scale analytics and AI transformation in South Asia. Prior to True Digital, Pedro worked at Axiata Group as group vice president of data analytics. Pedro, welcome and thank you for taking the time to speak with us today.
Pedro Uria Recio: Thank you very much. I’m very happy and honored to be part of this podcast.
Matthias Kamber: Pedro, can you talk us through your journey into technology and what drew you into the AI space?
Pedro Uria Recio: My education is in engineering, and that is when I first became familiar with AI, particularly with neural networks, which we used for handwriting recognition back in the ’90s. Then I started my career working in research and development in France and in Hong Kong. Later, when I was studying for my MBA from the University of Chicago Booth School of Business, I gained an entirely new perspective of my understanding of analytics and AI, which is the business perspective—for example, how to use analytics in order to improve marketing campaigns, to improve financial results, etc. After earning my MBA, I joined McKinsey & Company, where we started working with the first analytics applications at scale in business, in the early 2010s, and those applications were mainly around customer experience and marketing. Then I joined Axiata and became the new vice president of data analytics, and then became the chief data analytics and AI officer at True Digital.
Matthias Kamber: There is a lot of hype about generative AI since ChatGPT was opened to the public late last year. Can you tell us what generative AI is?
Pedro Uria Recio: Generative AI is causing a lot of global excitement. Corporations are engaging in pilots, and private equity and venture capital are increasingly funding generative AI. The media loves it, developers are embracing it, researchers are focusing on it. So there is a huge boom in generative AI.
Many of the technologies behind generative AI were actually developed many years ago, such as neural networks, but the newer technologies that are powering ChatGPT are things such as generative adversarial networks, autoencoders, foundational models, and a lot of newer technologies that have been developed over the past ten years. There is an explosion in technical approaches for generative AI.
And it is not only ChatGPT either; there are many exciting tools out there for human language recognition or generation, to program code, to generate all your images, video, etc. So ChatGPT is just one of the many tools out there.
And in general, when it comes to business, generative AI can do two things. On the one hand, it can enable automation by giving human tasks to software. And on the other hand, it can enable augmentation by providing humans with advanced tools and techniques to work more efficiently. These are the two things that generative AI can do.
Matthias Kamber: More importantly, could you tell us the key business applications for generative AI? Can you give us some examples?
Pedro Uria Recio: There are a lot of examples. Generative AI has versatile applications that span across the core functions of a business—such as marketing, sales, product development, supply chain operations, and engineering—and also across the support functions, such as IT, finance, strategy, legal, HR, and recruitment.
There are many start-ups out there focusing on a specific [application] for these verticals. For example, in marketing, generative AI can write marketing and sales content in the form of text, images, and videos, and it can also create product and user guides, or help create certain chapters of them. And it can also analyze customer feedback by summarizing and extracting themes or sentiments from the text of a conversation, such as a real conversation with a customer.
In terms of R&D, generative AI can help engineers make 3-D designs and prototypes, such as prototypes of industrial pieces that are later printed in 3-D.
For development and coding, which is where I am mainly focused, generative AI can help write code and documentation, and it can help debug code. When a code is failing, you don’t always know how to solve it, so generative AI can help you; it can propose new versions of the same code so that it runs better or faster. There are many applications for programmers and data scientists to use generative AI. In fact, some of the models of ChatGPT are specifically for programming.
I’m also particularly amazed by applications in support functions. For example, in legal, generative AI could draft or help you review a legal document. When you receive a new contract, generative AI can highlight aspects of that contract that could be potentially problematic for you or that you should pay attention to. So it can have many applications in legal and HR.
Matthias Kamber: Generative AI’s applications to business look impressive. It seems like businesses will need to adapt how they operate to use this new technology profoundly.
Pedro Uria Recio: Yes, you are right that they need to adapt. And businesses can adapt to use generative AI in three ways. The first way will be the easiest one and the third the most complex. They can start by using off-the-shelf generative AI applications, such as the products of the start-ups I mentioned, of which ChatGPT is just one of them. And I would guess that most people listening to this podcast have already started using some of them.
The second thing they can do is connect these applications like ChatGPT to their business processes through APIs, or they can even run more open models on their corporate systems. This is actually how ChatGPT makes money, by charging businesses to connect to their APIs so that businesses can use ChatGPT, for example, for customer service.
And the third approach, which is the most advanced one, is that businesses can adapt the generative AI models to their business context. They can train, for example, the generative AI with their company’s product catalog or the client data. And this can involve advanced prompting, prompting the AI with more advanced or specific questions. But most importantly, and this is a little bit more difficult, it can involve different levels of fine-tuning the AI models so that the AI model knows what your products are, what customers you have, who they are, what they care about, what is the segmentation of every customer, etc.
And something important to keep in mind when migrating or adopting AI is that having been successful in adopting data analytics is actually a requirement to implementing generative AI in an organization. So organizations that have not leveraged analytics at scale well are much worse positioned to implement AI, and they risk being left behind as competitors take advantage of the technology. Before you can implement AI, you first have to implement analytics.
And another thing that is quite important is one of my very hard-coded beliefs based on my experience: winning companies in implementing and scaling AI are not defined by the sophistication of what they implement but by their consistency in combining strengths across strategy, processes, and people.
Matthias Kamber: Based on these three implementation options, what new skill sets and capabilities will organizations need to seek in leaders to maximize their opportunities?
Pedro Uria Recio: Let’s start with the type of jobs to be done in order to implement generative AI, because the leaders are going to be the people supervising the shops. So the skills leaders need depend largely on what these new tasks and activities are going to be to implement generative AI. And businesses must develop a number of capabilities to be successful.
First, AI is an application that is running live, or running all the time in a business; it’s not running in batches that are triggered by a person. Therefore, businesses must implement live operations, or what is called Live Ops. It’s a way of managing an application that is running live, running all the time. Businesses must implement live operations to monitor and maintain AI models, so you need people for monitoring the AI system and the maintenance of the system.
Second, businesses also must develop new talent capabilities, such as familiarity with deep learning frameworks and machine learning and engineering skill sets for large, complex models. In most companies, particularly in emerging markets, where I am, most companies use very straightforward AI algorithms, such as logistic regressions, but those are simple algorithms, and those algorithms are not going to be useful anymore. We are going to the next level, which is deep learning, generative AI. So now you need to hire people who can work with complex algorithms and models, and you need to refresh your machine learning development toolkit and your processes and your talent for that.
The third thing is that businesses also must adopt tools to adjust and fine-tune these default AI models. Like generative AI, there is a default version of it, but your business is probably going to need something that is fine-tuned to its own needs. And you need engineers to adapt the AI to the context of your business.
And then, fourth, businesses should also do an architectural design of how they’re going to implement AI. There are many options to choose from, and they can choose which of the different components are going to be developed in-house and which are going to be outsourced. And then they have to evolve the technology stack accordingly, and that means a lot of jobs for data infrastructure people.
And finally, change management is going to become very, very important. A lot of people are going to be working in change management to help their organizations adapt to AI. You will need knowledge translators, people who can explain generative AI to businesspeople, how to use prompts, etc. And you will also need what is called a human in the loop, because AI shouldn’t run by itself. As I mentioned before, AI is going to be running live, but these applications make mistakes and someone has to be supervising them and be in charge. We call that person the human in the loop. Successful business leaders are people who know these new jobs, because AI is now going to be your new colleague or your new subordinate. So successful business leaders are people who understand how the new game is played.
Matthias Kamber: What is the new kind of leader who will be able to manage this complexity? What skills do they require?
Pedro Uria Recio: Generative AI is going to require new approaches to management and leadership, and the most important is determining who is going to be leading this change. The leader of AI in an organization has to be at the C-level. There is a study from a few years ago from McKinsey & Company about the top factor that differentiates winners from laggards in analytics. The top factor cited in the research was the presence of a tech-savvy C-level champion, who may or may not have the title of chief data officer or chief analytics officer but is the person who is driving the adoption. And, as I said earlier, the next evolution of analytics adoption is AI adoption, so having a C-level champion is also fundamental for AI adoption.
A C-level champion requires a number of personality traits and skills. To me, the most important one is, by far, a willingness to learn and experiment with new approaches that work alongside intelligent machines and algorithms, and this person’s ability to instill this culture of experimentation in the organization. This person also needs to have a strong understanding of the company’s business processes and goals, because that person is driving applications of AI into the business. He or she needs to be a visionary. And being data-driven and having some experience with AI and machine learning obviously helps. Last, the ability to effectively communicate technical concepts to nontechnical stakeholders is also a very important quality.
Matthias Kamber: Generative AI raises several regulatory, legal, and ethical questions for businesses and society. For example, how can ethical guidelines be established to ensure accountability and transparency in developing and using these systems?
Pedro Uria Recio: Generative AI is absolutely raising several regulatory, legal, and ethical questions for businesses and society, and there are questions that will have to be answered. We probably don’t know how to answer all of them now, but we will need to find solutions and answers as a society. Number one is who owns the output generated by AI? Who owns an essay that is written by AI? And who is liable for any error or damage that is caused by it? Also, how can user data be protected, how can the privacy of data be protected, and how can user data be integrated to train AI systems?
For example, a few weeks ago, ChatGPT announced that they will not be using client data to train the default model of ChatGPT. And that is fundamental and very important. That basically means that when you are talking to ChatGPT and asking questions that are relevant to you, ChatGPT is not going to learn information about you that it might tell to other people. Or when you connect your product catalog or customer database to ChatGPT through an API in your business, the default model of ChatGPT is not going to learn things about your customers. That protection of privacy is important.
And there are other questions: How can these AI systems be unbiased? How can they avoid situations or discrimination of different kinds? How can we create ethical guidelines across businesses?
And finally, accountability and transparency are important. So when an AI system is making a decision, it’s making a recommendation. Can we know the reasons why this recommendation is made? Is it transparent for a human auditor who wants to know this recommendation? What are the reasons the AI system came to that conclusion? These are very important questions, and we need to find a consensus about how to handle this.
Matthias Kamber: What is expected of businesses and leaders in this sphere, and do companies have the leaders they need to meet those expectations?
Pedro Uria Recio: Business leaders must first refresh their ethical framework to include many of these aspects that I have just mentioned related to AI: regulated compliance, data risk, explainability or interpretability of AI recommendations or decisions, and also data sharing, transparency, and reputational aspects. For example, if there is an appearance of offensive content or bias in the content generated by AI, there has to be a new ethical framework that is relevant to each business about how to handle these things, and this is one of the things business leaders need to do.
Matthias Kamber: How do you see the job market changing due to generative AI?
Pedro Uria Recio: Generative AI, and AI in general, is going to automate a lot of tasks that used to be done by people related to their jobs, some of which I’ve already mentioned. But generative AI is also going to create new opportunities and new jobs, and the adoption of generative AI is going to lead to changes in the job market because some skills will be in demand and others will not be. And this change is going to happen much faster than we think; it’s going to happen over the next few years.
So what is going to happen? Jobs involving tasks in front of a keyboard, such as programming, data science, marketing, and secretarial activities, are going to be much more susceptible to automation. We will see a decrease in demand for these jobs and therefore a decrease in salaries, which will be problematic for a lot of people. But conversely, jobs requiring interpersonal skills such as creativity, problem-solving, and emotional intelligence—for example, jobs in consulting, sales, and people management, those jobs that are related to people—are less likely to be automated and might see an increase in demand and therefore an increase in salaries. And we also have the new jobs required to actually implement AI. And those jobs are going to be very much in demand, and the people working in those areas will have very competitive salaries.
In 2019, I did a TED Talk about the impact of AI in the workplace called “AI will make the workplace more human, not less.” And I talked about these changes in jobs, how those jobs that are more related to interpersonal skills, more “human,” will be less automated. And those jobs that are more repeatable, more about “ brute intelligence,” are going to be more likely to be automated. So when I did this TED Talk in 2019, I never could have imagined this would happen so fast. And, as I said, it’s going to come really, really fast—in the next few years.
Matthias Kamber: Looking ahead, what advice would you give to leaders wanting to help their organizations meet their strategic AI goals in the future?
Pedro Uria Recio: If there is only one thing that I would like business leaders listening to this podcast to remember, it is to keep learning. AI is going to bring a lot of changes to your businesses, to your jobs, and it is your ability to learn fast and experiment fast that is going to make the difference, that is going to make you a winner. So as a leader, that is my top recommendation.
As for the business, there are a few best practices. For example, Accenture recently published a report identifying five differentiating factors between those companies that are winning in AI implementation and those that are not, and I totally agree with them. The first one is the existence of a champion that is leading in AI, as I talked about earlier. The second one is investing heavily in talent, and I think we can all agree on that. Number three is industrializing AI tools and things to create a strong AI core. Number four is designing AI responsibly. We have also talked about that, about all the technical, ethical, and regulatory issues. And number five is prioritizing long- and short-term investing, not only focusing on quick wins but also investing in platforms that are going to help implement more transformational changes in the business down the road.
Matthias Kamber: This is very fascinating. Pedro, thank you so much for taking the time to speak with us today.
Pedro Uria Recio: Thank you very much, Matthias.
Thanks for listening to the Heidrick & Struggles Leadership Podcast. To make sure you don’t miss more future-shaping ideas and conversations, please subscribe to our channel on the podcast app. And if you’re listening via LinkedIn, Twitter, or YouTube, why not share this with your connections? Until next time.
About the interviewer
Matthias Kamber (mkamber@heidrick.com) is a principal in Heidrick & Struggles’ Zurich office and a member of the global Technology Practice.