IBM, a global technology firm, foresees a future in which small language models will make artificial intelligence (AI) more accessible to organisations.
AI and automation are disruptive technologies that reimagine or redesign business processes.
This can help companies achieve significant productivity improvements, Paul Burton, general manager of Asia-Pacific for IBM, told the Bangkok Post.
This approach not only enhances competitiveness, but also allows for the creation of market differentiation, enabling companies to take the lead in their respective industries, he said.
"The most successful companies are those that simultaneously invest in technology, strategy and business process optimisation to capitalise on the transformative power of AI," said Mr Burton.
Spending on AI systems in the region is predicted to skyrocket from US$20 billion last year to $49.2 billion in 2026, according to IT research firm IDC.
Mr Burton said AI is going to augment jobs, but that human augmenting by computer is better than a computer by itself.
ALTERNATIVE MODEL
He said large language models (LLMs) offer immense capabilities, but require a high level of energy consumption and cost.
With models such as GPT-4 boasting trillions of parameters, the computational power required to train, tune and utilise these models is staggering, often amounting to hundreds of millions of dollars annually.
This limitation sparked the emergence of small language models (SLMs), said Mr Burton. With a parameter size of around 10 billion, it offers an alternative to users.
SLMs help democratise AI as they can be run at a lower cost on more attainable hardware, empowering more amateurs and institutions to study, train and improve their existing models, he said.
SLMs can also be run on smaller devices, allowing more sophisticated AI in scenarios such as edge computing devices and the Internet of Things.
At 13 billion parameters, IBM Watsonx Granite models are an example of SLMs that organisations can leverage to ensure ethical and responsible AI use and regulatory compliance, said Mr Burton.
Earlier this month, IBM expanded its AI software availability to 92 countries in the Amazon Web Services marketplace.
As SLMs focus on performing specific knowledge fields or use cases, this specialisation allows for better performance and efficiency in tasks such as code-converting customer service, document translation and summarisation.
SLMs are finding applications across various industries, ranging from healthcare to wealth management, he said.
Mr Burton said he expects more examples and applications will emerge in the next 6-12 months as the technology continues to develop at a fast pace.
These smaller models empower businesses and users to train the models themselves, even on personal devices such as laptops, he said.
MORE INVESTMENT NEEDED
As more hyperscale data centres are developed in Thailand, that digital infrastructure is driving advances in productivity, GDP and wealth, said Mr Burton.
However, simply deploying data centres is inadequate to generate value, as the marginal utility of additional infrastructure is zero without meaningful utilisation, he said.
To fully leverage digital infrastructure, complementary technologies and strategies such as hybrid cloud, cybersecurity, AI and their applications are essential, said Mr Burton.
"At IBM, we're investing in hybrid cloud to help companies leverage their technology investments," he said.
"We're investing in a $500-million venture fund to invest in AI firms, and we continue to look for other opportunities to help where it makes sense."
Nearly two-thirds of global commerce is related to digital technology.
Mr Burton said companies and government are pouring vast sums into digital transformation initiatives, an estimated $6.8 trillion between 2020 and 2023.
In Thailand, technology spending as a percentage of GDP is probably 0.7- 0.8%, while such investment in Singapore is 5%, he said.
In developed countries, the floor seems to be about 2% of GDP.
"Increasing technology spending while improving business processes, the regulatory environment and education can lead to significant economic growth," said Mr Burton.
For Thailand to become a trillion-dollar economy, he said it should prioritise education and skills, investing in technology to enable business process.
Mr Burton said AI is going to augment jobs, but human augmenting by computer is better than a computer by itself.
"Thailand needs to train people to collaborate and train new people entering the workforce, as the country that collaborates effectively with computers will emerge as the victor," he said.
