This is a guest post by David R. Hardoon, managing director of Aboitiz Data Innovation
If anything, the past few years have shown that artificial intelligence (AI) is more than just a buzzword. AI has successfully redefined the way that businesses work. From finance to healthcare, various industries have shown tangible outcomes from using AI as a part of their solutions, reiterating the immense possibilities this technology can provide for the world.
As more people continue to experience the benefits of utilising technologies to ensure businesses are future-ready, there is no doubt in my mind that AI will only continue to grow in importance in 2022 and beyond. This mirrors the sentiment of Kearney’s survey respondents, where more than 70% of them see AI as crucial to Southeast Asia’s future.
That said, the reality is that there is currently a gap between the number of people who favour AI adoption to be accelerated, and those that are truly able to adopt and operationalise it to its fullest potential, with over 80% of the region still in the early stage of AI adoption.
To close the gap and drive progress in the region, businesses must direct their attention to the accurate aspects of operationalising AI. There are a few common misperceptions that could be stalling the progress of AI adoption that often land companies in a catch-22 situation. So, let’s do some myth-busting.
Myth #1: You need an AI strategy
It may seem paradoxical, but in order to operationalise AI, businesses have to first forget their AI strategy. Instead, focus on your business objectives, and use AI as a tool to help achieve those goals. It is important to think of AI as an enabler, while the goals should guide how it is being used and implemented.
For example, in the Philippines, it took Republic Cement almost a month to measure the compressive strength of cement, which is crucial to the acceptance of the product for concrete mixes used in constructing vertical structures.
To transform the way the industry worked, we used a tool that predicts cement quality based on historical data to help operators and quality managers optimise concentrations of raw materials and save time. This resulted in better resource management and increased operational efficiency.
At the same time, we also improved the quality of cement products in terms of consistency and reduced CO2 emissions. Indeed, when businesses focus on their end goals and use AI as a “catalyst” to reach the destination, they can produce quality results.
Myth #2: There is a “perfect” solution and timing for when AI comes in
It is not always possible to know if a solution is “right” if you do not try it out. With human behaviour and technology constantly evolving, the AI solution you designed today based on a specific set of data and models may become obsolete by the time you decide to launch it. It is important not to be obsessed with perfection and wait for governance or frameworks to be in place before operationalising an AI solution.
To address the challenge of fostering financial inclusion for the unbanked population and providing fair access to banking and financial services for micro SMEs like mom-and-pop stores, UnionBank maintained a razor-sharp focus on these goals and evaluated the available data to find innovative solutions that were compliant with the bank’s risk appetite.
For instance, looking at power usage data in a restaurant to assess potential profitability for the bank may not be the best model to evaluate loan applications, but the key is to constantly work on improving the AI model until the goal is realised. Other data points and appropriate considerations can be used to refine the existing model. With consumer behaviour and market trends constantly changing, finding the “perfect” solution or timing could result in missed opportunities.
Sometimes, it is about operationalising AI solutions in identified quick wins considered as “low-hanging fruits” which build small steps to achieve game-changing solutions down the road. This can help gain buy-in for future AI projects to accelerate the adoption process.
Myth #3: Everything we do with DSAI should be kept under lock and key
AI transformations do take a lot of time and effort, and I can understand why most companies feel that knowledge gained during the process should be kept proprietary. However, I truly believe that sharing is caring. Through sharing, we can support the industry as a whole to progress, as we can mutually learn from each other on possibilities and challenges.
It is also important to note that though AI frameworks and models can be identical, data will never be the same as it has to be adapted based on the objectives of each individual business. By sharing case studies, we can short-circuit the sometimes arduous process of innovation and create more relevant and wonderful solutions.
With its myriad possibilities, AI has been trumpeted as a saviour to the problems of tomorrow. Hence, it is not surprising that the world is plagued by these myths as people believe that we have to be careful about its application, from planning an AI strategy to finding the “right” solution and protecting the knowledge gained from multiple trials.
However, it is exactly because AI presents so many opportunities that businesses need to focus on their end goal, share the knowledge gained with others to enable progress across industries, and have a ‘just do it’ attitude to try, try and try again because there isn’t a perfect plan.