The focus on strategic artificial intelligence (AI) for return on investment, agentic AI and rightsizing AI models are key trends that will drive AI's maturity for mass adoption by businesses next year, says IBM Thailand.
A unified AI framework is needed to ensure effective management and governance, as well as a human-centric AI approach, noted the company.
All these trends should spur investment from financial services, insurance, manufacturing, healthcare and universities, according to IBM.
"AI is rapidly transitioning from experimental trials to strategic implementation projects," said Anothai Wettayakorn, managing director of IBM Thailand.
In 2024, AI was increasingly infused into many organisations via hyper-experimental projects. However, only 5-10% of AI experiments have progressed to full deployment, noted the firm.
As the focus on AI-driven revenue takes centre stage, organisations will embrace the strategic AI approach in 2025, prioritising projects based on feasibility and business impact, he said.
The challenge is how to scale AI through use cases that maximise revenue opportunities and return on investment, said Mr Anothai.
Businesses will select high-impact and higher-risk use cases to gain greater return on investment and comparative advantage, shifting from low-risk, smaller use cases, he said.
Mr Anothai said IBM used AI transformation in 2023 to unlock US$1.5 billion in cost savings and productivity within two years. The amount is expected to reach $3 billion.
Another trend is rightsizing AI models, which requires significantly less training data and generates a smaller carbon footprint than large language models.
"Smaller, specialised open-source models are emerging as powerful alternatives for many AI applications," he said. "Purpose-built models designed for local languages and simpler computational tasks will be in high demand."
Mr Anothai also expects unified AI frameworks to ensure management and governance, as evolving regulations, diverse needs and responsible AI drive the needs for visibility, governance and seamless AI integration.
"Unified AI is a framework that helps businesses with multiple AI projects across different departments mitigate risks related to compliance, data privacy, legal issues, intellectual property and more," he said.
This helps eliminate risks from shadow IT, referring to the use of unapproved AI tools, software or platforms by employees or teams within an organisation.
Mr Anothai said IBM has a platform for unified AI, which aligns with the European Union's AI Act.
As AI regulations are expected to emerge in many countries, including Thailand, unified AI will help organisations prepare for future regulations, he said.
The rise of agentic AI is another key trend for the future of work, where AI agents autonomously execute tasks, collaborate with human workers and drive value across the business.
Agentic AI has the potential to achieve significant gains in operational efficiency, customer experience and decision-making, said Mr Anothai.
However, organisations must establish internal guardrails and regularly evaluate underlying models to ensure ethical and responsible use.
Lastly, the human-centric AI approach shifts AI from productivity tools to enhancing human experiences and capabilities. This approach will help employees automate routine tasks, boost creativity and unlock new opportunities, according to IBM.
By focusing on empathetic AI design, organisations can build stronger customer relationships and brand loyalty, he said.
Mr Anothai cited APAC AI Outlook 2025, commissioned by IBM and conducted by research firm Ecosystem, which found out of 76 respondents, the primary focus of AI investment for Thai organisations is back-office business process automation (29%), optimisation of IT functions (18%) and sales automation and customer lifecycle management (16%).
Key drivers of AI adoption are advances that make it more accessible (42%), environmental pressures (41%) and pressure from customers (39%). The challenges for AI adoption are vendor lock-ins (41%), lack of tools and platforms for developing AI models (34%) and the high cost of implementation and solutions (34%).
Nearly 60% of 518 surveyed organisations across Asia-Pacific anticipate realising the benefits of their AI investments within 2-5 years. Only 11% expect returns within two years.
IBM plans to help develop the AI Change Agent programme with three universities. The change agents are AI creators and lead innovation in the AI projects of businesses.