AI Skills / AI news for Malaysia
From the archive · Source report date 23 May 2025
Malaysia's farm AI gap is not only hardware. UPM says it needs local algorithm builders
Associate Professor Dr Christopher Teh Boon Sung framed the missing capability plainly: agriculture needs people who understand the field problem and can build or modify the software used to solve it.
In brief
- UPM academic Christopher Teh said many agricultural practitioners use existing software, while fewer people can build or modify it for local needs.[1]
- His work in agrometeorology and greenhouse-gas measurement illustrates why agricultural knowledge and data technology have to be combined.[1]
- A later UPM smart-greenhouse project showed the surrounding work: IoT, automated irrigation, remote monitoring and a mobile-accessible control system. It remained a pilot, not a national rollout.[2]
UPM identified a local software-building gap; it did not claim every farm needs a custom AI model
Malaysia's agricultural AI discussion often starts with drones, sensors or computer vision. UPM's Christopher Teh put the skills question one layer deeper: who can change the software when a farm's crop, weather, soil or operating practice does not match the assumptions built into an existing tool?[1]
UPM reported that Teh originally wanted to study computer science or computer engineering before entering agriculture. He later combined both interests and argued that local agriculture needs more people able to develop or adapt software, not only use whatever is already available.[1]
The gap appears when imported software meets a local farm
A ready-made platform may still be useful, but it arrives with choices about data fields, thresholds, crop conditions and recommended actions. When those choices do not fit a Malaysian farm, someone must understand both the agricultural consequence and the software change required. That is the capability gap Teh described.[1]
The distinction is practical. A user can enter data and read a dashboard. A builder can inspect how the result was produced, test it against field measurements, alter the workflow and document when the system should not be trusted. Agriculture needs both roles, but they are not interchangeable.[1]
Agricultural knowledge is part of the model, not an optional add-on
UPM said Teh works in agrometeorology and greenhouse-gas emissions measurement. Those fields depend on instruments and data, but the numbers only become useful when someone understands weather, soil, crops, measurement quality and the decisions a farmer can realistically make.[1]
That is why a generic coding course is not enough. A useful agricultural project should begin with a defined farm problem, record how the data was collected, compare the model with field reality and show whether the result improves cost, yield, labour, resilience or environmental performance.[1]
A smart greenhouse shows the work that sits around an algorithm
UPM's later AgriSMART Lestari report documented a mushroom greenhouse with an IoT ecosystem, automated irrigation and remote monitoring. The control system could be accessed through a mobile application, while the cultivation process used mushroom blocks made from kenaf biomass supplied through an external collaboration.[2]
The report called it a pilot and said it had the potential to become a reference model. That boundary matters. A pilot can show that sensors, controls and cultivation practices work together in one setting; it does not by itself prove the same economics, reliability or support model across Malaysian farms.[2]
The education pathway has to combine field competence with software judgment
UPM had already introduced a four-year smart agriculture programme with coursework and an industrial mode. Its stated aim was to produce technologists able to manage sustainable smart-agriculture systems, and the university reported an initial intake of 20 students for the 2022/2023 session.[3]
Teh's argument raises the bar for programmes like this. Graduates should be able to use agricultural technology, but also trace a result back to its data, recognise a weak assumption, work with developers and validate a change in the field. The valuable graduate is the bridge between farm reality and software behaviour.[1][3]
Why Malaysia should care
Malaysia can buy sensors, platforms and models, but local value depends on people who can connect those tools to Malaysian crops, weather, farm practices and operating constraints. UPM's evidence points to a combined skills problem: agronomy, data, software and field validation must meet in the same workflow.
Agriculture students
Technology literacy becomes more valuable when it is attached to a real crop, measurement and operating decision.[1][3]
Practical move: Build one field project that documents the dataset, model limits and practical result.
What Malaysians can do now
- Start with one farm decision and define the measurement that would make it better.
- Document the local data, operating limits and human fallback before automating the workflow.
- Judge a model by field performance and farmer value, not by technical novelty alone.
What we still do not know
The skills direction is clear, while demand, curriculum depth and farm-level economics remain unmeasured.
- How many Malaysian agriculture roles currently require software or model-building skills rather than tool operation.
- How deeply local programmes teach programming, data engineering, model validation and long-term system maintenance.
- Which smart-farming pilots have produced independently measured gains that remain economical beyond one site or funding cycle.
Sources
- 1.Minat Algoritma Jadi Kerjaya Dalam Bidang Pertanian Universiti Putra Malaysia, 23 May 2025
- 2.UPM Vice Chancellor Launches AgriSMART Lestari, IKP Smart Agriculture Initiative Institute of Plantation Studies, Universiti Putra Malaysia, 14 May 2026
- 3.UPM develops 3u1i study programme in smart agriculture Universiti Putra Malaysia, 6 October 2022
- 4.Directory of Academic Staff: Christopher Teh Boon Sung Faculty of Agriculture, Universiti Putra Malaysia
