The business case for AI

Artificial intelligence is the hottest topic in technology. It has captured the imagination of tech specialists and the wider public as a solution that promises to generate innovative solutions for consumers and productivity gains for businesses.

AI is not new, but driving the current wave of enthusiasm is the rise of generative AI (GenAI), which came onto the scene with the launch of ChatGPT in 2022.

What makes GenAI such a significant development from previous iterations of AI is its ability to create new content and its accessibility, which allows a broader range of non-specialists to work on AI-related projects. The question going forward is whether this new technology can provide real business value.

How businesses are taking advantage of AI was the subject of a panel discussion at the HSBC 11th Annual China Conference, where experts with practical experience in the field shared their insight on the subject.

Use cases – consumer vs enterprise

The panel started its discussion on the current balance of use cases, between consumer usage and the ways that businesses are integrating AI into their operations.

At the level of individuals, GenAI has really taken off, with people using solutions like ChatGPT to generate content and virtual assistants to help organise their day-to-day activities. But at the business level, there has been slower progress – especially for companies operating in regulated industries.

“In the enterprise world, the adoption rate really depends on the industry a company operates in,” said Michael Yung, Strategic Advisor, APAC Industry Technology Group, Google Cloud. “You really need to understand what the technology can and cannot do, and you really need to take care that the customer can accept its implementation.”

The finance industry is an example, where front office jobs will likely rely on human beings for some time to come. Regulators are still not comfortable with AI solutions selling products, such as insurance, even though an automated solution might be less likely to make a mistake than a real person.

This means that a lot of AI integration is taking place in the middle and back office of financial institutions. An insurance salesperson might not be able to remember all the information relevant to their job, so they could ask a non-client facing AI solution to answer a question, with the information subsequently delivered to the customer in person.

But the panel said that it is only a matter of time before regulators develop a framework on how they are going to govern AI. And when that happens, a wide of range of new solutions will hit the market, delivering significant productivity gains. The panel agreed that it is important for technologies to keep an open door for regulators so that policymakers can fully understand what the technology can do.

Monetisation channels

When it comes to monetisation, AI has different models depending on the scale and ambitions of each company. For the very largest businesses, an API-based model might work best, as there is potential for developers to make solutions that can reach an enormous scale. The downside for an API producer is that it has to wait until the community creates something that takes off – potentially a long-term process that requires a large reserve of capital.

Smaller companies can adopt a model of customising APIs, which are then licensed to end users. Although this requires fewer engineers, the downside is that it is hard to get recurring revenues from one-off license sales.

Another model, appropriate for the smaller companies is to focus directly on applications. One company that has adopted this approach is Fano (Fano Labs Limited), which specialises in speech and natural language processing to help companies in areas like customer services and compliance.

“An application is tied to a use case, and a customer can see whether it will help them make money or save money,” said Miles Wen, Co-Founder and CEO, Fano Labs Limited. “Our revenues tend to have higher stickiness as customers get continuous value from our product.”

The main drawback for this model is that focusing on one particular application can be very niche and the total pool of potential revenues will be smaller than in other business models.

AI in China

Home to many of the world’s largest technology companies, there are high hopes that Chinese businesses will be able to fully utilise AI. The panel pointed out that the country faces several challenges that will need to be overcome.

For a start, although China has an enormous hardware industry, its software industry is relatively small. Software-as-a-service (SaaS) is not a popular operating model in China, making it hard to sell software. As a result, the economics of building a large model in the country do not make sense. What Chinese companies are doing instead is building products for very specific use cases – especially in the industrial space.

Another challenge is hardware, as US export restrictions have cut China’s access to high-end chips. There is a serious attempt to make domestic alternatives, but it is a challenge to deliver a product that can compete with the global industry leaders in a market that is relatively small and where customers are very price sensitive. That said, there will be use cases where domestically made chips will economically fit the particular business needs for individual companies.

Finding business models that work, as well as getting the necessary hardware will be essential for Chinese success in the world of AI. And overcoming these challenges could have wide-ranging consequences, as AI is one of the next-generation technologies that could become a new growth driver for the Chinese economy. The development of Chinese AI will therefore be a major technology theme in the years to come.