2025 is predicted to be the year when artificial intelligence (AI) will make major invasions into the mainstream. In fact, IBM believes that Asia-Pacific companies are expected to maximize the impact of AI investment beyond AI experiments.
In a conversation with the Manila Times, managing partner Pia Azarcon, managing partner at IBM Philippines, discusses the state of AI in 2025 and the opportunities and challenges it presents to APAC and the Philippines as well. Azarcon oversees the use of platforms such as IBM’s own Watson X and other large-scale language models from third parties.
The Manila Times (TMT): What is the current state of AI adoption worldwide? Especially in ASEAN and the Philippines?
Azarcon: In the context of your question, let’s first look at what happened with AI in 2024. Basically, there have been a lot of experiments, especially for my clients. There were many proof of concept (POCs) – pocket initiatives across each organization.
The CEO adds that he was extremely enthusiastic about technology-driven initiatives. They were also serious about trying to match these AI efforts to the outcome of their business. I really have to say that most CEOs really want to explore how AI can adapt to their business.
They did their best to land the basic abilities. The AI initiative was very new, but CEOs realized that in order to do it across the enterprise, they need to scale very expensive and have to match business values. That was where all the experiments came from.
TMT: When you say clients, are you talking about global or ASEAN clients?
Azarcon: Domestic clients. Here in the Philippines! But if I’m talking about ASEAN, I’ll take the page out of an AI outlook that actually reflects what I said.
In 2024 there were tentative adoption and many experiments. And as we move on to 2025, the question now is how to land this. How do you use the experimental results? How can you scale them so that you can drive revenue, reduce costs, optimize costs, and increase efficiency in your operational processes? At the same time, regulatory concerns must also be addressed.
TMT: I would like to argue that IBM is a global company has successfully adopted AI in more advanced economies that could replicate in countries like the Philippines and regions like ASEAN. .
Azarcon: Absolutely! When we started, there were a few use cases. Since then, use cases have expanded significantly across business processes, businesses and operations, including finance, HR, and even pay, customer service and BPO.
When we talk about agent AI, we use AI to support and double with human agents to provide work. Currently, there are many use cases for Generator AI. So it’s time to move on to 2025 and propose many generative AI projects for production.
Returning to business value again, the number one area must be strategic AI, as it is centered around five areas. And when it comes to strategic, AI initiatives must impact the business and align with the client’s organizational strategy.
And secondly, before it becomes something like shooting the moon. POC here, such poc. In 2025, businesses will be right size. You can probably see smaller, targeted projects using smaller open source models, but before they were all large language models.
Take this into consideration. First, small language models are more cost-effective. When you are still training people to become proficient in using generating AI, you cannot expect the whole organization to clarify it at once. So you start with a small team. There is a second reason. You’re the right size. Small teams need less training and even less data. Not all companies have the complete data available for generating AI. So it makes sense to be smaller and can be expanded later.
The third is to see companies that leverage many generative AI tools, especially those that have governance and orchestration tools. So, at least in the projects they are starting, they have better visibility to all stakeholders. It also makes it easier to integrate with small projects that can be scaled later. So you see them coming together, uniting them, and working on security concerns at the same time.
There is a fourth agent AI. It means an agent, helping and double the human worker to make work faster and accelerate turnaround time. Therefore, the agent AI used with IBM actually helps you redefine your workflow.
In other words, combining these two efforts to split tasks for humans and robotic agents will not only make the task faster, but also improve decision-making and in turn. Drives everything, especially the efficiency of operation.
And the fifth calls it human-centered innovation. If I put it in the right context, it’s about consumers like you and me. Generated AI helps you improve your experience. Consuming services with AI innovations generated there will improve the customer experience or human experience.
Therefore, across these five concerns, we can see that in 2025, the 2024 experiment will lead to changes in the more grounded implementation of this year’s generation AI.
In summary, if I’m an organization and want to employ AI, I’ll definitely reevaluate the data capabilities. I appoint a champion of my organization to promote governance that promotes my fundamental capabilities. It drives the selection and nomination of data sources that allow me to dominate my experiments as well as initiatives that are strategically aligned. It’s an organizational roadmap, so you don’t just pay for what you spend on using and generating AI, but you actually make money.
Competitors are pinching everyone’s heels, so at some point they’ll say they’ll actually have to do it, even if they don’t want to. They will be left behind.
