Breaking down the AI ecosystem

This article is part of our Opinions section.

AI continues to dominate the headlines this year. But as the initial hype begins to subside, the conversation has started to shift beyond the merely aspirational to a more purpose-driven discussion about the actual.

To engage in a more concrete dialogue about AI, we need a deeper understanding of the AI ecosystem. Today, policymakers distinguish between AI developers and deployers while enterprises consider the “build vs buy” dilemma. However, these simplistic binaries do not capture the complexities of this rapidly evolving industry and what is happening on the ground. With so much of the media attention focused on tech heavyweights occupying long verticals, we tend to forget that the real merit of AI is in its potential to transform smaller players, allowing them to operate at much larger scales and to punch above their weight.

We are still in the early days of the rise of AI. As Ali Golshan, Founder of Gretel AI, aptly puts it: “If you compare a mature market to a mature tree, we’re just at the trunk.” Yet, the broad outlines of the growing ecosystem are already beginning to take shape. Setting aside the AI chip industry, the key components of the AI ecosystem are outlined in the box below.

Whether you are a regulator developing AI frameworks, a policymaker aiming to grow your domestic industry, or a business leader crafting your enterprise AI strategy, a deeper look into the AI ecosystem is invaluable for informed decision-making.

AI regulations and liability

There is a general global consensus on core AI principles like transparency, fairness and privacy. But as discussions advance toward specific AI safeguards and regulations, there needs to be a conversation about how regulatory obligations and responsibilities are allocated.

The shared responsibility framework for cloud computing where obligations are distributed between data controllers and processors sets a useful precedent. Fundamentally, obligations should fall to the entities best positioned to discharge them. For instance, disclosure requirements for model training and testing should fall to the model developers, and not an application developer or integrator.

In addition, obligations and liability need to be allocated in a manner that reflects operational and marketplace realities. If one specific segment bears an overly large proportion of the compliance or liability burden, it will compromise the balance of the ecosystem and hinder growth.

Industry development

Governments worldwide are eager to cultivate their domestic AI ecosystems. Japan recently launched its Generative AI Accelerator Challenge (GENIAC) to enhance its platform model development capabilities by providing access to computational resources and facilitating collaboration with data holders. Singapore and Indonesia are collaborating to build their SEALION large language model, envisioned to support the growing regional AI industry. South Korea just announced plans to invest USD$527 million to spur AI innovation.

But growing the ecosystem is not just about pouring in investments – a luxury not all countries can afford. Policymakers need to identify the key gaps in their ecosystem.

For example, if an ecosystem suffers from poor data management and quality, startup grants and subsidies will not be effective. Instead, efforts should focus on improving data sharing and establishing data quality best practices.

Policymakers also need a sense of where their domestic players might have a comparative and competitive advantage to allocate resources efficiently. For instance, building a large foundation model requires significant funding, computing resource, AI talent and time, and may not be the most efficient use of limited government resources.

Enterprise AI strategy

Businesses today face increasing and competing demands on their technology budgets. With growing cloud requirements and escalating cybersecurity needs, the excitement over AI and generative AI solutions can seem like just the tip of a very expensive iceberg.

Business leaders need a robust enterprise AI strategy to avoid being left behind or rendered obsolete by AI advancements while also preventing overinvestment. This involves not only considering how AI can streamline processes and reduce costs, but also how it can transform the business landscape altogether by enabling new products, services and business models.

Leaders must think strategically about their company’s role in the broader AI ecosystem and identify what differentiates them from competitors.

These may still be the early days of the rise of the AI industry, but as the Chinese proverb goes: “Qian ren zai shu, hou ren cheng liang.” (Those before us must plant the trees so that those after can enjoy the shade.) Planning and preparation need to take place now if you do not want to be left behind, and strategic and targeted AI investment grounded in an understanding of the evolving ecosystem will make the most difference. 

Worth a read

• The imperative of making a career in the data centre industry attractive
• AWS plugs another $9 billion into Singapore and launches new AI program
• Ghostbusters proton packs in real life