Approximately 32 per cent of organisations surveyed in the Asia Pacific region have shown their dedication to investing in generative AI technologies, according to a recent report by global intelligence firm International Data Corporation (IDC). 

And 38 per cent of these respondents are actively exploring potential use cases to implement using generative AI, it said.

'The IDC Survey Spotlight: What Is the Attitude of Asia/Pacific Enterprises Toward Generative AI Adoption and Application?' reveals that digital-first enterprises are keen on utilising it as a strategic tool to enhance enterprise intelligence and drive efficiencies in various functions including marketing, sales, customer care, research & development, design, manufacturing, supply chain, and finance.

In the Asia Pacific region, generative AI in knowledge management is leveraged for access and search across large repositories of information of different types of images, documents, voice, and other formats across an enterprise.

Another significant use case involves code generation, where application programmers utilise generative AI to streamline code creation, optimisation, completion, testing, and debugging, resulting in improved productivity and code quality.

Additionally, generative AI is leveraged in marketing automation and customer-facing roles, allowing marketers to generate highly tailored and search engine-optimised content, as well as develop conversational applications.

“Generative AI has the potential to reimagine the organisational landscape in a completely new way. However, the inherent complexities and risks around implementing the same needs to be carefully assessed,” says Deepika Giri, Head of Research, Big Data & AI, IDC Asia/Pacific including Japan (APJ) Research.

“Generative AI technology is also largely in its early stages, as vendors are unable to fully address the privacy, security, accuracy, copyright, bias and misuse concerns around this groundbreaking technology,” Giri adds.

In the race to capitalise on this technological advancement, various vendors are vying to lead the way. They range from hyperscalers and cloud service providers offering Model-as-a-Service (MaaS) solutions, AI engineering companies with point solutions, to specialist storage companies who want to sell the infrastructure that can host these solutions.

Additionally, investment firms are actively seeking substantial returns by investing in this technology. The demand for extensive data to test large language models (LLMs) has given rise to companies offering synthetic training data that can be leveraged for training purposes, to address issues such as the use of sensitive data, bias, etc.

The practical implementation of generative AI can be as straightforward as acquiring off-the-shelf solutions for marketing, customer care, and code generation. In the market, there is a wide range of vendors who have integrated GenAI capabilities into their offerings across these domains.

Alternatively, large language models (LLMs) can be adopted and trained or fine-tuned for specific use cases, which can be a complex and resource-intensive process involving significant computing and energy costs.

Prompt engineering provides a simplified way to train the model by writing “natural language” type queries to elicit the right response. Another technique, called prompt tuning, offers a simpler way to train the model without requiring complete retraining or parameter adjustments, thereby reducing computational requirements. This approach strikes a balance between the extreme approaches

Regardless of the chosen approach to utilise generative AI technology, there is an inherent cost associated with the underlying infrastructure, as the model is compute heavy. The price of computing is either in the form of an upfront investment to set up the datacenter or built into the price of the MAAS offering - there is no escaping it.

The study also highlights the growing global concerns regarding the application of generative AI, which currently lacks comprehensive governance at this stage. Regulatory bodies are facing pressure to address issues related to data privacy and security, intellectual property rights, and the potential misuse of AI-generated content. Governments at best are trying to adhere to existing policies around these areas rather than formulate new ones.

The Indian government has chosen not to regulate AI, as it believes that AI is a catalyst for the digital economy, and imposing strict laws at this stage could hinder innovation and research. The Japanese government has set up a council that intends to promote AI technology to keep up with the global interest in the subject. The Cyberspace Administration of China (CAC) has unveiled security assessments and the impact of Generative AI services before being launched in public.

Despite the aforementioned concerns, there is currently no specific legislation in place for generative AI in the Asia Pacific region. This is because stringent regulation could be seen as an obstacle to fostering innovation in a progressive digital economy. Different countries in the region are at varying stages of developing their approaches to AI regulation, resulting in a lack of uniformity in their perspectives.