In fact, it can weaken ESG returns, inflate energy costs, add unnecessary OpEx, and deliver almost no measurable business value.
问题 is not AI.
问题 is how organisations procure and deploy AI.
There is a clear difference between AI bought for hype and AI bought for business value.
AI for hype means provisioning a large model for every small workflow, even when the workflow does not need that level of compute. It looks modern in a board slide, but behind the numbers it is inefficient, expensive, and wasteful.
Using AI properly means matching the right capability to the right business need.
从 the 财务 desk at AINNA, I see AI 智能体 as part of asset management, not just productivity 工具.
An AI agent should allocate the right “asset class” to each task. A low-cost, 轻量模型 handles repetitive work like data entry or invoice matching. A heavy, GPU-intensive model should only be deployed when the work is high-value, such as complex forecasting or compliance review.
But if we force one expensive LLM to handle every task, from the smallest to the biggest, then GPU and energy costs stay high around the clock.
That is not smart automation.
That is just poor asset utilization with a fashionable label.
The future of AI 面向马来西亚SME should not be about deploying the largest model everywhere.
It should be about building efficient 系统 where every task gets the right level of intelligence, the right level of compute, and a clear return on investment.
AI should trim the 损益表, not bloat it.
That is where real financial innovation begins, and where AINNA delivers measurable value to Malaysian SMEs.
Official Cross-Network Resource
