Dig deeper to strike AI gold

THE artificial intelligence (AI) race just got a whole lot more interesting. DeepSeek, a Chinese AI firm, recently unveiled its R1 model, developed at a fraction of the cost of its US rivals yet with similar impressive capabilities.

For perspective, it costs less than US$6mil to develop the Deepseek R1 Model compared with an estimated US$40mil to US$100mil for the Chat GPT-4 Model.

This AI “Sputnik moment” challenges the assumption that only companies with massive resources can compete in AI, effectively throwing a wrench into the US’s grand AI plans.

Its massive US$500bil Stargate Project – by OpenAI, Oracle, SoftBank, MGX and in partnership with other tech heavyweights – is now faced with tough questions.

Moreover, a Berkeley research team recently replicated a smaller scale DeepSeek model at just US$30, and Alibaba’s Model Qwen2.5 claimed to outperform the R1 Model entirely.

While the United States and China are locked in a fierce battle for AI dominance, Malaysia needs to carve out its own path.

We cannot simply mimic these giants; we need to leverage our unique strengths and position ourselves strategically in this evolving landscape. As the Malay proverb goes, ukur baju di badan sendiri (measure the clothes on your own body).

The data gold rush

The AI race isn’t just about fancy algorithms and futuristic robots. While this may be obvious, the AI race is about making data useful – turning raw information into powerful tools that can be accessed anytime, anywhere capable of transforming how humans and economies work.

Just like the California Gold Rush, while everyone was scrambling for gold, the savvy ones were selling the shovels and pickaxes.

Today, data is the “gold”, and the “shovels” are the algorithms, processing chips, software and data centres that make AI possible. It’s a race to mine, refine and present data in its most useful form. And just like in the gold rush, those supplying the shovels – the tools and infrastructure for AI – will ultimately strike it rich.

Fortunately, Malaysia has recognised the importance of AI – establishing the National AI Office (NAIO) to drive the country’s AI agenda.

We are, in theory, trying to balance between these two superpowers to get a slice of the pie with our National Semiconductor Strategy and the recently announced specialised AI Working Groups under NAIO.

But we need to go further.

When it comes to AI, we cannot just focus on shiny new models. We need to think about the whole ecosystem – who’s developing the tech, who’s providing the data, and who’s ultimately using it.

Just like in the gold rush, different players have different roles, and we need different strategies to manage each of them.

This means taking a close look at how everyone’s operating, their ethics and their values to our national growth.

We need to find that sweet spot between encouraging innovation and protecting our interests, including setting clear rules for those who might not have our best interests at heart, while nurturing local champions in different parts of that ecosystem.

I have started an initial mapping of the AI ecosystem according to the value chain with preliminary suggestions, potentially for a possible PhD research paper.

I’m happy to discuss with any interested policymakers.

Here’s what Malaysia needs to consider:

1 Data is the beginning and the end for AI models.

Recognise that data is the fuel for the AI revolution. Securing our data must take a multi-dimensional approach. AI models are trained on our data that is available on the Internet.

We need to safeguard the data that belongs to Malaysians, today and in the future. This means we need a data sovereignty framework for data sharing and ownership, an enhanced Personal Data Protection Act, clear guidelines to the Data Sharing Act and standards on data ownership among others.

We must also control what happens to our data once it’s in the AI models. We cannot let the data be manipulated by the model owners.

There are already reports of DeepSeek models censuring any criticism of China; Google dropping its pledge not to use AI for weapons and surveillance; and AI models losing control due to data input by users (refer to the incident of an Israeli AI Chatbot developed for zionist propaganda going rogue and becoming pro-Palestinian).

In response, we must be regulating AI algorithms and ensuring source transparency for responses is made a feature of AI regulation.

2 A shovel to call our own.

We need to get serious about building our own tools and tech that power AI, while taking into account our existing capabilities.

This requires boosting our local capacity to build the hardware needed including the chips, data centres and processors that fuel AI.

Think homegrown companies like Globetronics Technology Bhd and Inari Amertron Bhd becoming their own AI chipmakers.

Concurrently, we need to invest and nurture our local AI talent and support the growth of Malaysian AI companies that are capable of developing and deploying cutting-edge AI models right here in Malaysia –though we must acknowledge it’s a steep hill to climb at the moment.

3 Regional frameworks to regulate Big Tech.

Big Tech often leads the development of AI, possessing vast influence and resources that trumps the majority of countries’ economies. We should look into developing and streamlining regional frameworks to ensure that there are uniform standards to curb the risk of Big Tech abuses.

Ultimately, the AI revolution is here, and Malaysia has a choice to make.

Will we be passive consumers of technology, or will we seize the opportunity to become innovators and leaders?

To succeed, we need a clear-eyed vision of our current capabilities and our future aspirations. Let’s build an AI ecosystem that empowers Malaysians, protects our interests and positions us for success in this exciting new era.