AI and the future of Technology and AI work: what changes, what stays human

Key facts

• In US data published in March 2026, Anthropic found its AI model, Claude, was being used for 75% of the tasks listed for computer programmers, more than for any other job. US usage data, not Malaysian demand^[1].
• On the ILO's 2025 index, software developers score 0.53 out of 1 (gradient 3) and web and multimedia developers 0.60 (gradient 4, the most exposed band). Global generative-AI task-exposure estimates; not a Malaysian forecast; not a job-security rating^[2].
• Employers in the World Economic Forum (WEF) survey put big data specialists, AI and machine learning specialists and software developers among the fastest-growing jobs to 2030. Global employer expectation, published 2025^[3].
• In Malaysia, the Human Resources Minister said in August 2025 that about 30,000 ICT jobs are expected to be significantly impacted within three to five years, based on a TalentCorp study. "Impacted" does not mean lost^[4].

Technology work met AI first, because AI tools are software and programmers adopted them early. Most of the measured evidence is from the United States, and the Malaysian figures are projections. Use this page to find questions worth asking a working developer.

For routes in, SPM subjects and courses, start with Technology & AI after SPM.

How to read this page

Observed means someone measured it. Projected means someone estimated the future, so we say who, when and for where. Projections are often wrong.

The scores come from the International Labour Organization (ILO). They estimate how much of an occupation's current tasks generative AI (tools that write, summarise, code or make images from a prompt) could potentially do, from 0 to 1^[2]. "Not exposed", the lowest band, is a technical label: most of the occupation's tasks scored consistently low for what generative AI can do today^[2]. It does not mean AI will never change the work, and it says nothing about robots or Malaysian employers. Exposure is not job loss either: the ILO says that depends on adoption and on whether workers get the chance to learn the tools^[2].

What is already happening (observed)

• Coding is the biggest use in one AI product's data. In Anthropic's February 2025 data (Claude.ai conversations sorted into US job categories), computer and maths tasks were 37.2% of conversations, more than any other group^[5]. Anthropic usage data, not Malaysian demand.
• AI already writes a lot of code. In April 2025, 79% of conversations on Anthropic's coding tool, Claude Code, were automation (the AI doing a task with little input from the person), against 49% on its general chat app^[6]. Anthropic usage data, probably skewed towards early adopters.
• Use still trails what is possible. Anthropic estimated in March 2026 that language models could in principle help with 94% of tasks in US computer and maths jobs, but Claude was used for 33% of them^[1]. US data.
• Young US workers: an early signal. Stanford researchers found employment of 22 to 25 year olds in the most AI-exposed US jobs was about 19% below where it would have been had it kept pace with less-exposed jobs (August 2026)^[7]. Anthropic found job-finding rates for 22 to 25 year olds entering the most exposed US jobs fell about 14% compared with 2022, with no such fall for older workers^[1]. Both are US data, and both teams call it early evidence, not proof that AI caused it.
• Malaysia. We did not find a current Malaysia-specific study that measures how AI is changing tech jobs here.

What people project, and who says so

• Tech employers worldwide. 99% of employers in information technology services expected AI to transform their business by 2030. They said people mainly do 41% of tasks today, technology 22% and the two together 38%, and expected 26%, 34% and 41% by 2030. Software developers and AI and machine learning specialists topped their growing roles; data entry and accounting clerks were shrinking. Global employer expectation, published 2025^[3].
• Employers in Malaysia. 82% expected AI to transform their business by 2030, close to the global 86%, and put AI and machine learning specialists first among roles they expect to grow. Expectation of employers in Malaysia in the WEF's global survey, published 2025^[3].
• TalentCorp, for Malaysia (reported November 2024). TalentCorp, the Ministry of Human Resources' talent agency, studied 10 sectors in 2024. Of about 3.5 million jobs, it rated 18% (620,000) as highly impacted, meaning they may merge with other roles or be displaced, and 53% as still relevant but needing upskilling^[8]. The minister later put 30,000 of the highly impacted jobs in ICT (August 2025)^[4]. We read news reports of the study, not the report itself.

Task by task

These cards are our reading of the sources for a classroom exercise, not a forecast. A practitioner may disagree, and their reasons are worth hearing.

Write code for a new app feature

• What a person does: decides what to build, reads what was written and owns the bugs.
• What AI might assist: drafting and completing code.
• What this source does not show: whether Malaysian firms hire fewer juniors because of it.

Build a web page or app screen

• What a person does: checks it works for real users, on cheap phones and slow connections.
• What AI might assist: turning a sketch into working code.
• What this source does not show: how Malaysian clients will price this work.

Clean a dataset and report what it shows

• What a person does: chooses the question and catches results that make no sense.
• What AI might assist: writing the analysis code and first charts.
• What this source does not show: which roles Malaysian employers will merge or split.

Keep servers and accounts running

• What a person does: decides what to do when something breaks, and answers for it.
• What AI might assist: writing scripts and reading error logs.
• What this source does not show: how often AI-made changes cause outages.

Help a colleague whose laptop will not connect

• What a person does: handles the odd case and the stressed user, sometimes in person.
• What AI might assist: chatbots answering routine questions.
• What this source does not show: how many support jobs in Malaysia change.

Work out what a client actually needs

• What a person does: runs the conversation and notices what was left unsaid.
• What AI might assist: drafting a requirements document.
• What this source does not show: anything about trust, which no score measures.

New tasks that are appearing

• Reviewing and testing code an AI wrote, and explaining it to a teammate.
• Writing clear specifications so an AI agent builds the right thing.
• Checking AI systems and AI-written code for errors, bias and security holes. Employers list security management specialists and information security analysts among the fastest-growing jobs (global employer expectation, published 2025)^[3].
• Protecting the data that AI tools touch.

This list is our reading of the sources above, not an official one.

What stays human: hypotheses to test

These are practitioner hypotheses to test, not findings. The sources do not show that any of these tasks will stay unchanged.

• Who decides what is worth building, and for whom?
• Who takes responsibility when software fails someone?
• Who understands a user's real situation, such as a stall owner who wants a simpler way to take orders?
• Which tech work happens in real buildings, with cables and ladders?

Ask a developer which of these hold in their job today.

What this means for you

• Learn to code, and learn to check code. If AI takes more of the routine typing, beginners may be expected to review, test and explain work earlier. That is a reasonable reading of the US evidence, not a certainty for Malaysia.
• Build something for a real person. A small tool that a relative, club or shop actually uses teaches you how to find out what people need.
• Pair tech with another field. Health, logistics, finance and energy all need people who understand both the work and the tools. Employers in Malaysia put supply chain and logistics specialists second among roles they expect to grow (expectation of employers in malaysia in the wef's global survey, published 2025)^[3].
• Use AI as a tutor, not a ghostwriter. Read what research says about learning with AI before you lean on it for coursework.
• Watch the entry-level signal. The US data on young workers is early. Watch Malaysian graduate employment figures, TalentCorp's next studies, and the National AI Action Plan 2026-2030, led by AI Malaysia since its launch on 28 July 2026^[9].

Your next step

1. On Day 2, choose the Technology and AI taster: build a small tool with an AI assistant, then list every step you had to fix and why. If your cohort runs four labs instead, read the Build and Engineer lab brief and use its software developer row.
2. Use the Job X-ray kit on "software developer", starting from its tasks on e-MASCO(opens in a new tab), and ask a practitioner which of your labels they would change.

About the evidence

ILO scores (a global generative-AI task-exposure estimate; not a Malaysian forecast; not a job-security rating), on a 0 to 1 scale with the ILO's band^[2]:

• Web and multimedia developers 0.60, gradient 4
• Database designers and administrators 0.57, gradient 3
• Systems administrators 0.57, gradient 3
• Mathematicians, actuaries and statisticians 0.56, gradient 3
• Software developers 0.53, gradient 3
• Systems analysts 0.49, gradient 2
• ICT user support technicians 0.47, gradient 2, with tasks that vary widely
• ICT installers and servicers 0.24, not exposed

The bands run from "not exposed" and "minimal exposure" up to gradients 1 to 4, the most exposed^[2]. The ILO notes that highly digital jobs also have a strong capacity to develop new tasks as the tools change^[2]. It built its scores from a survey of 1,640 workers in Poland, rounds of discussion with international experts, and an AI assistant built on that input. Every task counts equally, and the ILO says local factors its global study does not cover can speed up or limit the change^[2]. ILO occupation groups do not map one-to-one onto Malaysian job titles, and we do not claim a direct mapping.

WEF figures are expectations from a survey of more than 1,000 employers across 55 economies, published January 2025^[3]. Anthropic and Stanford figures are usage data or US data, not Malaysian demand. The method behind the task cards is in Tasks, not job titles.

For facilitators

Use the task cards as a starting list for the Day 2 AI task X-ray and ask the practitioner to correct them before learners use them. Each learner supports one card with a source and challenges one assumption. Stop any conclusion that a job is "safe" or "dead": the evidence here does not support either.