Get Started

Filling the APAC AI Gap: Insights From IDC InfoBrief

Scaling AI Frederic Catherine

For the third consecutive year, Dataiku has commissioned an IDC InfoBrief that deep dives into the state of AI in APAC,  “Artificial Intelligence Practices: Driving Business Differentiation in the Digital-First Economy.” This InfoBrief presents the consolidated findings of two IDC AI adoption surveys completed in 2021 to track the development of regional organizations’ core capabilities in managing their AI investments. 

Key highlights of the InfoBrief include: 

  • Across the board, adoption momentum has slowed down (compared to 2019), with 39% of businesses in APAC already using AI in their business operations.
  • Most leading use cases are industry specific (i.e., risk management in banking, store and channel operations in retail, or drug safety in life sciences). Although more complex in nature, they have a greater potential for driving more business value.  
  • While AI capabilities are improving among the leaders and organizations in APAC, a new divide is emerging: Early adopters are seeing the most growth currently, as this segment has already captured low-hanging opportunities and are now implementing use cases of larger scope and complexity. The lack of momentum among laggards and the continued improvement from early adopters suggests an increasing self-perpetuating divide. 
  • Those with AI capabilities will have a clear advantage in the current digital- economy. 

Understanding the Current Context

To understand the challenges, we must first understand where we stand. AI adoption remains steady at 39% (as of 2021) in the APAC region. However, it should be noted that momentum has slowed compared to 2019 — possibly as a result of COVID-19, inflation, and geopolitical uncertainties. 

These early adopters are expected to increase their spending by 34% on average — as they are now entering a new phase, where they scale their operations thanks to the success of their early implementation of AI models. Additionally, their focus is shifting from internal-focused goals, such as improving employee productivity and business agility (top adoption drivers in 2020), to more external-focused goals, such as driving product and service differentiation and improving customer experience (2021). 

This shift is a result of the positive strides towards a digital-first economy. We know that the economy will accelerate on its digital destiny with at least 65% of APAC’s GDP digitized by 2022, driving U.S. $1.2 trillion of direct digital transformation  investments from 2020-2023. As a result, a clear divide between laggards and early adopters is arising. 

What Can Organizations Do?

To ensure AI implementations deliver consistent returns, organizations need to design an enterprise-wide strategy to coordinate their AI investment and strengthen their four core capabilities: data, people, technology, and process.

However, IDC shows that over 60% of APAC organizations do not have an enterprise-wide strategy to coordinate their AI investments. Here’s how they can get started across each of the aforementioned four pillars:

1. Build Data Capabilities

While we see an improvement in AI capabilities in the region, a closer look shows that, across regions and industries, the levels of data readiness have only seen a slight increase — with an average score from 2.6 to 2.8 out of 5. Data capabilities is a challenge, but it should not refrain companies from starting or scaling their AI strategy. Organizations need quality data, not perfect data; the risk being that organizations chase perfect data to build models instead of rapidly iterating, which leads to fewer and more risky deployments with bootleg data. As we’ll see in the next section, as businesses accelerate their data capabilities and start their AI journeys, demand for talent is only going to grow, making people harder to hire for and, simultaneously, driving up salaries.

2. Build People Capabilities

The core challenge of a successful AI strategy lies in the people. The InfoBrief shows that only 42% of APAC organizations employ the right skilled personnel for their AI functions.

The biggest deterrents to AI innovation in people's capabilities are the resistance to change and lack of skill set:  

  • Skilled talent is more than just additional data scientists, they should also cover areas from solution architects and software developers to MLOps engineers. 
  • Amongst organizations in APAC, only 53% have hired AI solution architects and 43% plan to or have hired MLOps engineers in 12 months.

Undoubtedly, sourcing for the right talent can be difficult and retaining even more so. Therefore, organizations need to focus on promoting better data culture, fostering collaboration, and lowering the learning curve. Commonly, reducing the skill gap is a feat that takes time. Providing the right tools and programs with automation and low-/no-code features will not only lower the skill barrier but also allow AI professionals to focus on higher, value-add activities.

Data scientists are a hot commodity in the job market, resulting in higher tendencies to job hop, so businesses must build a culture that marries data with its business functions. Using a program with collaborative features allows different personas to be involved. By making the use of data and AI an everyday behavior, organizations can more easily empower all people (including the business) in a central place, accelerate the time it takes to deliver AI projects from months to days, and seamlessly govern the lifecycles of all AI projects. As we’ll see in the next section, the scarcity of this talent means fewer individuals with the requisite experience to build and manage the process.

3. Build Process Capabilities

Implementing process capabilities allows the institutionalization of metrics and processes needed for data operations, model operations, and associated business operations, ensuring the discipline required for consistent and superior business performance.

Organizations are still struggling with operationalization — over 75% of models are still not going into production as of 2021 — indicating that APAC organizations must improve their ability to scale model delivery. As organizations continue to scale AI initiatives with more models, data sources, and stakeholders, it becomes more important for the strengthening of business process capabilities. 

Appointing an AI model steward to oversee the governance protocols and practices of the data to model lifecycle will ensure all necessary documents and reviews are in place. It will also help teams navigate any foreseeable challenges in ensuring governance in the model development and management processes with the least friction.  

4. Build Technology Capabilities 

Organizations can leverage technology platforms to tackle challenges when it comes to people and processes.

From a people standpoint:

  • Automation and low-/-no-code features will lower the skill barrier. AI professionals will be more productive and can focus on higher, value-add activities.
  • With collaborative features, different personas involved will become more engaged and less resistant to change initiatives.

From a process perspective, leveraging technology platforms is critical for model development and operations. Businesses need to have a platform to orchestrate workflows and automate training cycles, ensuring the timeliness and consistency of model delivery and updates.

Machine learning (ML) model retraining is fundamental to ensure that models are constantly up to date while minimizing manual interventions and optimizing for reliability and monitoring. As AI adoption continues to expand in APAC, the retraining cycle also continues to become shorter.

How fast does this need to happen? InfoBrief shows that over 63% of APAC organizations investing in AI need to rebuild their models at least weekly. These advancements highlight the increased importance of updating and retraining and the introduction of integrated platforms for continued model development and operations. 

An overwhelming 98% of Asia Pacific organizations that have invested in AI are currently using commercial ML software platforms, and for good reason! Commercial platforms offer greater compatibility, more agility, and easier access to all stakeholders in the company.

Looking Ahead

As the adoption of AI expands in APAC, businesses must develop digital capabilities to generate additional revenues and stay ahead of competition. Improving data readiness, reducing the skills gap, operationalizing AI, and building tech capabilities/platforms will help businesses in the region to work towards achieving a common True North. To ensure AI investments and implementations reap consistent rewards, it is vital for businesses to leverage technology and opt for an integrated platform that enables everyone across the organization to be part of building AI within their ecosystems.

You May Also Like

Explaining AutoML: What It Is and How Dataiku Can Help

Read More

Succeed With AI at Scale With These New Year’s Resolutions Tips

Read More

5 Reasons Why Predictive Maintenance Is Overhyped

Read More

Fairness: The Bedrock of a Responsible AI Future

Read More