Pockets of success

The push towards AI has seen forward-looking organisations in Southeast Asia embrace innovative AI applications such as chatbots and recommendation engines, says Chua.

In Indonesia, developers at startup Kata use natural language processing to power an intelligent chatbot that speaks and understands the Indonesian language to improve customer engagement. Indonesian telco Indosat also uses AI and advanced analytics to glean unique customer insights based on the behavioural and location data of its users.

At Foodpanda, one of Singapore’s online food delivery services, machine learning algorithms do the humanly impossible.

“We use a very complex algorithm that self-learns from tens of thousands of operations and past orders, so that we can send our riders to the restaurant only when the food is ready”

Luc Andreani, Foodpanda Singapore

“The time required by restaurants to prepare food may vary on different days and times,” says Luc Andreani, managing director of Foodpanda Singapore. “We use a very complex algorithm that self-learns from tens of thousands of operations and past orders, so we can send our riders to the restaurant only when the food is ready.”

That particular algorithm has enabled Foodpanda to strike the fine balance between optimising the number of deliveries a rider makes and ensuring that food is delivered as freshly cooked as possible.

Foodpanda also uses another algorithm to ascertain the value of its compensation vouchers given to customers who have encountered poor service, and to analyse the redemption rates of those vouchers to determine if a bad experience has been converted into a good one.

In the financial services arena, both OCBC Bank in Singapore and Al Rajhi Bank in Malaysia are tapping AI for anti-money laundering to improve operational efficiency and accuracy in detecting suspicious transactions.

Unequal access to the internet

A prerequisite for more widespread AI adoption is the presence of digital infrastructure that can systematically collect and disseminate large quantities of data.

“The efficacy of machine learning algorithms is only as good as the data that is available,” says Manjunath Bhat, research director at Gartner.

While the digital revolution has reshaped the business world, Southeast Asia still lags in digital readiness. Unsurprisingly, digital infrastructure is not evenly distributed across the region’s 11 countries, with just over half of the region’s population using the internet.

Governments have a key role to play to further AI adoption in Southeast Asia. For example, governments can oversee regulations to accelerate open banking initiatives that can democratise access to data, while maintaining personal data privacy, says Bhat.

Among Southeast Asian nations, Philippines and Singapore lead the pack in open data implementation, followed by Indonesia, Malaysia and Thailand, according to the Open Data Barometer.

However, in line with global trends, progress on open data remains slow, according to an Open Data Barometer report, which found that only 7% of datasets in the East Asia and Pacific region were open. Key information that can be used by citizens to hold governments accountable, such as contracts, companies and spending data, remain closed.

The skills shortage is also a formidable challenge to AI adoption. With most of the new innovation in AI coming from startups in the ASEAN region, access to capital is no longer a challenge for these startups, but access to world-class talent is.

“There’s a skills shortage for AI globally and the small size of the region exacerbates the skills shortage further. The limited availability of AI skills and experience will mean a bidding war for talent in the coming years,” says Bhat.

ViSenze’s Tan adds: “Our universities need to do better with producing coders and software engineering graduates. There needs to be more students in computer engineering if Singapore is serious about AI.”

AI adoption still nascent

Successful AI adoption requires a complete understanding of the role of data and technology that can integrate findings from data collected over time.

Even with pockets of successful AI projects across the region, most businesses are still in the nascent stages of AI adoption, marked by gaps in the understanding and approach to AI technology, data and skills – as well as the lingering fear that AI will take away jobs.

“AI/machine learning is simply not an easy technology to apply – it requires a good understanding of the business problem, and which data and machine learning tools will address that problem,” says Chua.

Keith Lim, CEO and founder of Hearti Lab, a startup that focuses on AI in the insurance industry, says: “Key decision-makers in enterprises who are reviewing investments in AI often face resistance from internal stakeholders because they fear that jobs may become obsolete.

“This resistance can be overcome, when the enterprise understands that human employees are able to improve productivity with AI, through careful resource planning and allocation.”

While the quantity and variety of data captured by enterprises is growing exponentially, the ability to derive actionable insights from this data is also limited by the existence of data silos, with data fragmented across multiple systems.