AI can write, but can't feel

THE HANOI TIMES — After Typhoon Yagi, a journalist arrived at a flattened village in northern Vietnam. He found a man sitting on a cracked concrete pillar—one of the few pieces of his house that had not been swept away. Around him was nothing but bare earth and scattered bricks.

His wife and two children had died the night before.

The journalist introduced himself. The man nodded, didn’t move. He wasn’t crying. He had cried all night. Now he just stared at the space where his house used to be.

“I don’t even know where to start again,” he said. “I didn’t even have time to say goodbye.”

The journalist stayed half an hour. They talked. He listened. He tried to ask questions, but most of them felt useless. He took some notes, recorded a few quotes, and then left with a notebook full of words and a mind full of doubt.

Back at his desk, he tried to write the story. He erased every draft. He left the page blank for weeks. Only months later, on a talk show about disaster reporting, did he talk about it. "I couldn't bring myself to write the story," he said. "Not because I didn't want to, but because it pained me too much to get it right."

With a single click, artificial intelligence (AI) can now generate dozens of news stories. It can scan damage reports, pull data, and write neat summaries. But it will never sit next to someone who has just lost everything, wondering which part of their pain the public should see. It will never hesitate or feel guilty about not knowing what to say.

This is no longer a technical debate; it is a matter of identity of what journalism truly stands for, and what must be protected as technology becomes more capable than the people who once mastered it.

Take the Panama Papers, for instance. It wasn’t a machine that uncovered one of the biggest leaks of offshore financial dealings in history. It was a network of investigative journalists who followed leads, cross-checked records, and made judgment calls no algorithm could replicate. It was about intuition, persistence, and a deep sense of public accountability.

The future of journalism will not be defined by a fight between humans and machines. It will be shaped by whether humans can lead machines, not follow them.

There is no denying that AI is already changing the way newsrooms operate. At the Associated Press, the number of automated sports stories published per quarter increased from 300 to more than 4,000 after adopting AI systems. These tools have helped reduce production costs and improve speed, but not efficiency. Volume is not meaning. Journalism without the human element becomes hollow.

No machine has ever stood at the site of a collapsed house. No AI has ever heard the trembling in a survivor’s voice or understood the silence of a village after a storm. While AI can generate coherent text, it cannot carry grief, take responsibility, or show restraint. It cannot sit with someone’s story and decide which part should be shared and which part should remain sacred.

This is not about computing power. The Moravec Paradox reminds us that while machines excel at logic and data, they struggle with what humans do effortlessly: empathy, perception, and moral judgment. AI can recognize patterns in words, but it cannot grasp why a father might go back into floodwaters for a final search. It cannot sense the difference between tragedy and spectacle. Only a journalist can make that distinction.

At a recent forum on journalism and AI, Professor Nguyen Duc An made it clear: “Artificial intelligence can process language, but it cannot feel context, ethics, or human pain.” It can mimic tone, but not intention. It can generate a voice, but has no sense of presence. It never carries the burden of what it publishes.

AI can summarize damage reports and scan satellite images in seconds, but it does not hesitate before publishing a photo that could retraumatize a grieving family. It does not flinch when transcribing the account of a mother who lost her child. And it certainly does not stop to ask: “Should this part be left out?”

After Typhoon Yagi, AI could have produced articles filled with statistics and maps, but it could not have stood in the wreckage, listened to a man who lost his wife and children, and captured the depth of his silence. Choosing to tell that story or to hold it is an ethical act. And ethics cannot be programmed.

As AI becomes more advanced, the need for human oversight only grows. When Microsoft’s MSN published a story about racism and mistakenly used the wrong photo of a Black artist, the damage was immediate. No one blamed the algorithm. They blamed the newsroom.

Even more alarming is the rise of synthetic content. Deepfakes, cloned voices, and AI-generated articles are spreading rapidly. The ability to fabricate is becoming easier than the ability to verify. And without journalists to question, contextualize, and push back, misinformation will not just spread but dominate.

This is where human journalists matter most. Not by matching machines in speed, but by doing what machines cannot. They ask the hard questions, recognize nuance, bring ownership to the stories they tell, interpret events, and decide how truth should be told with clarity and with care.

Some may argue that AI will one day replace journalists. But AI does not walk through disaster zones or lose sleep over ethical dilemmas. It does not face threats for speaking truth to power or standing before a grieving father, wondering how to write a story that is both honest and kind.

The role of journalism is not just to report, but to witness. A journalist is not just a content generator, but a custodian of truth, of memory, of human dignity. In an age where information is abundant but meaning is scarce, this role is more important than ever.

The challenge ahead is not whether journalism can survive AI. It is whether it can survive without remembering why it matters. As machines get better at writing, journalists must get better at feeling because no algorithm can replicate what most defines journalism: the ability to care and the courage to choose how.