Concerns and implementations of AI in Malaysia’s digital landscape at the Malaysia Digital Tech Adoption Summit

• Suggestion to take notes from the EU AI Act 
• Successful AI implementation starts with a clear, structured approach

The Malaysia Digital Tech Adoption Summit on 12 Sept showcased the transformative power of AI across industries, from large-scale infrastructure needs to practical applications like fraud prevention. AI is already reshaping the way businesses operate, driving growth and efficiency.

As Malaysia positions itself as a hub for digital innovation, there’s growing recognition that AI is not just a future tool but a present-day catalyst for change. Hosting the summit reflects MDEC’s commitment to supporting the country’s development as a leading digital economy by supporting the country’s AI ecosystem and helping it to strengthen, therefore playing its part in creating a more competitive and resilient economy.

A panel, ‘Navigating AI Landscape’ was held as part of the summit.

“Today, big companies and hyperscalers are building 100-megawatt data centers over a period of eight months,” said Fahri Aminudin, resource director of GDS International, speaking at the 

GDS Holdings Ltd is a China-based developer and operator of high-performance data centers. It opened the Nusajaya Tech Park Data Center Campus in Johor in 2023.

Accompanying Fahri were Tiensoon Law, Deputy CEO of Innov8tif and Tan Aik Keong, CEO of Agmo Group Bhd, who shared insights into how their companies are navigating AI integration and driving innovation in their respective fields.

Moderated by Ts.Fadzli Abdul Wahit, Senior Vice President of MDEC, the panel delved into the diverse AI applications, strategies, and challenges faced by organizations striving to keep pace with this transformative technology. 

In GDS’s case, Fahri was referring to their role in the AI landscape. “We are preparing our infrastructure in terms of power, in terms of cooling, as well as the GPUs and CPUs for these hyperscalers.”

“In the DC (data centers) business today things have to be done fast as today, big companies and hyperscalers are building 100-megawatt data centers in over a period of eight months,” he said.

A hyperscaler is a company that operates large-scale data centers, providing massive computing resources typically through cloud platforms. These companies support extensive infrastructure for large-scale data processing, storage, and management.

Fahri also emphasized the necessity of power and cooling of today’s data centers.

“10 years back, in terms of rack cabinets for instance, it was just around six to seven kilowatt. Today, we're talking about a 100-kilowatt rack to process data to support AI,” he said.

"We're talking about 100 times faster, so we need to have sufficient power,” he added.

Implementing AI adoption throughout an organization

When asked for his view on what are some critical success factors in terms of implementing AI  adoption throughout an organization, Aik Keong said that there are several things that are needed.

“You need to start with identifying your goals and what are your problem statements,” he said. “Once you identify it, you could address the vulnerabilities or automate some processes."

He also brought up the importance of checking data availability and quality

“Sometimes you might not have the data with the relevant digital format,” he added.

Once the data has been collected, one needs to set up the right orders and identify if any customization is required. For example, if you need a legal model AI, you don’t have to use a large language model, as a small one would be more efficient.

Tiensoon agreed stating that “While AI is the biggest keyword of today, many of its real-world applications are built on smaller models thanks to machine learning and deep learning, not necessarily general AI applications that we are using today,” he said.

He also stated that with LLMs (large language models) earning a lot of attention, more people are becoming experienced AI users, which lowers the barrier of entry into developing AI.

“That's probably the reason why this year, we see a lot of companies like software companies and marketing agencies embracing generative AI and LLM to create a lot of usable applications,” he said.

Last but not least, Aik Keong also stressed the importance of a roadmap. “AI implementation to me is a lifelong journey because it evolves constantly, so you have to treat it like a product.”

As businesses increasingly look to AI to enhance operations, he stressed that successful AI implementation starts with a clear, structured approach. "To implement AI effectively, you first need to identify your objectives—what are the problems you're trying to solve? Next, ensure you have access to quality data, as the absence of relevant digital data can be a major hurdle. Once you’ve secured your data, you can determine whether customisation is needed. For example, domain-specific AI, like legal AI, may not require large models; a smaller model trained on your data could be more efficient."

Aik Keong emphasised that AI implementation is an iterative process: "After setting up, the next step is to appoint an AI lead to oversee training and adoption. Governance is also crucial, along with planning a clear roadmap for the next three years."

Agmo Group’s approach highlights the need for businesses to be both technologically and organisationally ready for AI, with a focus on long-term, structured development.

AI guidelines and addressing concerns

With AI rapidly growing and revolutionising industries globally, there are concerns raised in regards to the risks and challenges it poses in terms of both ethics and security.

While Innov8tif doesn’t have its own set of guidelines, Tiensoon said that they are using guidelines from AI Verify Foundation, a non-profit organisation established in June 2023 by the InfoCommunications Media Development Authority of Singapore (IMDA), with the aim to create international guidelines for AI’s responsible, secure, and innovative use.

“Some of its tools can help you test your data set, mostly the structured ones, and they'll be able to tell you whether they’re biased or fair,” he added.

He also pointed out a concern of AI regarding data privacy guidelines, to which he hoped that there would be clear guidelines when it comes to the usage of IC number and phone number across the industries.

“With the roll out of e-invoicing for all businesses till July next year, IC numbers are going to be circulated everywhere,” he said.

Aik Keong is aware of the concern and pointed out an example of an AI regulatory framework in the form of the EU AI Act which was implemented in August.

Firstly, an AI system has to follow regulatory compliance. 

“If your AI is a high-risk system, as in it negatively affects safety or fundamental rights, it has to go through this conformity assessment before it could even go into the market to ensure transparency,” Aik Keong said.

Secondly, for general purpose AI models, they’ll need to go through things regarding documentation; they’ll need to have a summary of its data training to ensure it isn’t in the high-risk category.

Thirdly, its standards are applied internationally (both EU and non-EU entities) as long as AI solutions are used in the EU, the Act applies.

With the fear of AI taking over jobs, this has led to a concern regarding the future workforce.

However, Aik Keong presents the counter-argument that people that don’t have AI skills won’t be replaced by AI but by those who use it.

According to him, Agmo is not only focused on offering AI solutions but also on preparing organizations for AI-driven futures. Through its AI Academy, the company helps businesses measure their technology readiness and provides tailored training programs to upskill employees. 

This holistic approach ensures that the company is not just adopting AI solutions but is also building the internal capacity to leverage it effectively.

Fahri on the other hand understood the concern that some people couldn't adapt and learn machine language, C-language, Python because “it's too late”.

“Therefore, we need to have more developments be it software, data centers, providing power, because of pay rate that data centers are offering is on a much higher level, so that's why we set up in Singapore, just to narrow the disparity of the income, which is what drives talent as well,” he said.