Editor’s Pick | By Edwin Wong, CEO, The Ledger Asia
There is a question I have been thinking about quite a lot lately. If an AI agent can monitor hundreds of news sources, identify an important development, research the subject, extract the relevant facts, write a story, generate a headline, prepare the SEO metadata, select a category and potentially send the article into a publishing workflow, what exactly are humans supposed to do? Or, to put it more provocatively, do we still need to work if agentic AI can work for us?
For anyone running a digital newsroom, this is no longer a science-fiction question. We have already moved beyond the stage where artificial intelligence is simply something sitting in a chat window waiting for us to type a prompt. The next phase is agentic AI. These systems do not simply answer questions. They can be given objectives, connected to tools and allowed to perform multiple tasks in sequence. In a newsroom, that changes the conversation completely.
An AI agent could wake up before the editor does. It could scan regulators, stock exchanges, government ministries, company announcements, financial markets and international news organisations. Another agent could decide which developments appear significant. Another could build a research dossier. Another could prepare a draft. Another could check the article against editorial rules. Yet another could prepare the headline, meta description, slug, tags and social-media copy. All before someone has finished the first coffee of the morning.
If that sounds slightly uncomfortable, I understand. But I do not think the right response is to ask whether AI is going to take our jobs. I think we should ask a different question: what part of our jobs was genuinely worth humans doing in the first place?
This may sound strange coming from someone in the media business, but a considerable amount of newsroom work is not really journalism. It is administration around journalism. Copying information from one system into another, checking websites repeatedly, cleaning documents, transcribing interviews, reformatting press releases, creating metadata, preparing summaries, tagging stories and monitoring whether something has changed are all necessary tasks, but they are not usually the reason people become journalists.
Very few journalists enter the profession because they dream of spending their careers copying text between browser tabs. If an AI agent can remove much of that work, I am not convinced we should be mourning its disappearance. I would rather have journalists spending their time asking better questions.
This is where I think the debate about artificial intelligence and employment sometimes goes wrong. We keep imagining a competition between human and machine. The machine becomes more capable, therefore the human becomes less valuable. I do not necessarily see it that way. I think AI is gradually pushing human work further up the value chain.
Consider breaking financial news. An agent might detect that Bank Negara Malaysia has released a new statement. It can retrieve the document immediately, extract GDP growth, inflation figures, monetary indicators and important quotations, compare those figures with the previous quarter and prepare a perfectly readable first draft. That is impressive. But there is still another question: so what?
That question is surprisingly difficult to automate. A figure can be factually important without being editorially important. GDP may have increased, but what part of the economy is actually driving it? Exports may have jumped, but is that sustainable? Inflation may have eased, but are households actually feeling better? A company can report record profits while quietly revealing something far more significant in one paragraph buried deep inside its filing.
The AI can find information. A good journalist needs to understand why that information matters. That distinction becomes increasingly important when machines can produce unlimited quantities of technically competent content.
AI gives us the ability to produce enormous amounts of content, but that does not mean we should. The internet already contains more information than any human being could possibly consume. If every newsroom deploys 100 AI agents and every agent produces 50 stories a day, we have not necessarily created better journalism. We may simply have created more noise.
The competitive advantage of a newsroom therefore starts to change. It is no longer simply about how much we can publish. It becomes a question of what we should publish. That is an editorial decision. Editorial judgement is ultimately about choosing what deserves attention, what does not, which questions need answering, whether the available evidence is strong enough and whether a story adds understanding or merely adds another URL to the internet.
In an AI-powered media environment, restraint may become almost as valuable as productivity. If production becomes easy, selection becomes more important.
There is another issue that matters even more: accountability. Suppose an autonomous news agent publishes something incorrect about a company. Who receives the lawyer’s letter? Certainly not the AI. Suppose it misinterprets an economic statistic. Who explains the mistake to readers? Not the model. Suppose it publishes an allegation that technically appeared in a source but lacked proper context. Who takes responsibility? Again, a human does.
That is why I do not believe a serious newsroom can simply become a room full of servers. Journalism is not merely an information-generation business. It is a trust business. Readers are not just consuming words. They are trusting that someone has exercised judgement before those words reached them. Technology can participate in the process, but it cannot inherit the responsibility.
There is also little value in pretending nothing will change. It will. Some newsroom functions will become dramatically smaller. One editor with a sophisticated network of AI agents may eventually perform work that previously required several people. A journalist who understands AI-assisted research may investigate subjects far faster than someone manually opening 40 browser tabs. Smaller publishers may suddenly gain capabilities previously available only to organisations with large editorial teams.
That is going to change newsroom economics. Jobs built mainly around repetitive information processing are likely to come under pressure. I do not think we should hide from that reality. But I also do not think the conclusion is that humans become unnecessary. The conclusion is that the definition of productive human work changes.
Imagine a digital newsroom a few years from now. An editor arrives in the morning and opens a dashboard. During the night, several AI agents have already been working. One monitored Asian financial markets. One watched regulatory announcements. One analysed corporate filings. One followed geopolitical developments. One reviewed technology and artificial-intelligence news. The system has identified perhaps 200 possible stories.
But instead of publishing all 200, it presents the editor with 15 recommendations. The editor starts asking questions. Why is this important? What is missing? Have we checked the original document? What happens if this trend continues? Who benefits? Who loses? Is everyone reporting the same angle? What can we tell our readers that they do not already know?
The agents go back to work. They research, calculate, retrieve, compare and draft. The editor decides. That looks less like replacing a journalist and more like giving a journalist a small digital research department. And that, to me, is where things become interesting.
So, back to the original question. If agentic AI works, do humans still need to work? Perhaps not in the way we used to. We may spend less time gathering information manually, less time formatting documents, less time performing repetitive research, less time writing routine updates and less time moving information from Point A to Point B.
But that does not mean there will be nothing left for us to do. Quite the opposite. It may mean that more of our working day is spent on the things machines still struggle with: judgement, context, curiosity, relationships, original reporting, scepticism, editorial taste and responsibility.
These are harder to measure than word count or publishing volume, but they are exactly what distinguish a newsroom from a content factory.
My bigger concern is not that AI will replace journalists. It is that journalists will start behaving like AI. Publishing whatever appears in the feed, rewriting what everyone else has already written, optimising everything for algorithms, measuring productivity by the number of stories produced and accepting information without asking enough questions.
If that happens, then perhaps AI should replace us, because we would have reduced journalism to precisely the kind of work machines are very good at doing.
The opportunity is to move in the opposite direction. Let machines handle more of the mechanical work so humans can become more human, not less. More curious. More analytical. More willing to challenge assumptions. More interested in understanding people. More prepared to say, “Something about this does not make sense.”
That sentence may turn out to be one of the most valuable skills in the AI newsroom.
I believe we are approaching a period when very small editorial teams will be capable of operating surprisingly sophisticated newsrooms. AI agents will continuously research. Workflows will automate production. Software will handle much of the operational machinery. Humans will sit above that system, directing it, questioning it and ultimately taking responsibility for what it produces.
So I do not think the future is a newsroom without people. I think it is a newsroom where fewer people can do substantially more, provided they know what is worth doing.
And perhaps that is the real question agentic AI is forcing all of us to confront. For generations, work has partly been defined by the amount of effort required to produce something. AI is beginning to separate effort from output. When producing something becomes almost effortless, value has to come from somewhere else.
In journalism, I believe that value will increasingly come from judgement.
AI can read almost everything. AI can summarise almost anything. AI can write extraordinarily quickly. But someone still has to decide what deserves to be said.
For now, I still want a human answering that question.
And yes, that means we still have work to do.
