Prime Time: The Debate With Ghosts

If an artificial station has a stance, let its human keeper come and defend it

By Mukhtar Alam

In the rush to appear modern, Pakistani prime-time television recently hosted an unusual debate. Two artificial characters, presented as female representatives of their respective AI stations, were interviewed on matters of public interest.

The segment was called innovation. In journalistic terms, it was a non-event. Technically it was fake, morally it was misleading, and journalistically it had zero news value.

To understand why, we must separate two very different uses of Artificial Intelligence in the news world.

The first use is AI as a tool. This is now common globally. Since the first experiments around 2018, newsrooms have used AI to reformat long texts into simple language, translate bulletins, generate illustrations where no picture exists, and make statistical projections. The global trend is that serious newsrooms are expanding this backend use while investing more in human investigation — because that is the only content AI cannot replicate. This use is honest. The journalist remains in control and accountable for the fact; AI only improves the presentation.

The second use is AI as a person. This is where the recent prime-time example falls. Here, AI was not used to present information; it was presented as a stakeholder giving its "version" and its analysis.

This is a fundamental confusion. An AI station cannot have a version. A version implies an interest, a risk, a bias, a profit, a loss, a fear, a hope. An artificial character has none of these. It says what it is prompted to say. This is not journalism; this is ventriloquism. Earlier, a ventriloquist at least showed his skill. Now the anchor hides his hand. He writes the prompt and makes the dummy speak his own wish. He asks what he wants to be answered, and the dummy answers what he wants to be heard.

What is the news value in interviewing someone who does not exist? An interview's worth lies not in its smoothness, but in the accountability, the field experience, the conscience and the reputation of the interviewee. A dummy has no past to be checked, no future to be worried about, no reputation to lose.

If an AI station does have a stance to offer, that stance belongs to its keeper — the human entrepreneur who built it, funds it and owns it. He has a physique, an address, a business interest, and a potential to deliver and to be held accountable. If we must know what an AI station thinks, let its natural owner come on screen and defend it. Let him face cross-questions. Ghosts cannot be interviewed on matters that affect living citizens; only humans with a body, a stake and an address can.

If this practice is not checked today, tomorrow's picture is clear and it is anarchy. We will have actors without a script and without a director. Every prime-time show will have its own stable of agreeable dummies. A dummy economist for the budget who will always praise the policy, a dummy analyst for foreign affairs who will always support the anchor's line, a dummy activist who will never ask a tough follow-up. No need to call a real professor, no need to face a real critic. Just type a prompt and the show is done. We will be generating and promoting fiction, myths and wishes as news.

And why give the dummy a gender, a pretty face and an attractive name at all? For spectacle. For eyeballs. There was a time when theatre had a message and spectacle had a limit. What we are seeing now is absurd theatre, where prime time — a trust earned over decades when viewers leave their families to listen to serious issues — is turned into time-pass.

One may think that advanced countries had prepared laws for this years ago. They had not. The EU passed its AI Act only in 2024 and its implementation for media is still underway, the US still has no federal law on AI in news, and the UK is still on voluntary guidelines. The virtual wave caught everyone unprepared. The difference is, they started a debate on disclosure and accountability. We started celebrating the fake.

Our own regulator is also unprepared. PEMRA's code, written in 2015, was made for a different era. Its net was designed to catch loud abuse, hate speech and vulgarity. It was never designed to catch a silent, well-dressed lie. Its monitoring is complaint-based and looks for the old sins, so a fake interview passes easily. No abuse, no shouting, just fiction spoken as fact.

But read its preamble: authentic information, public interest, public trust. Under its own spirit, this content is a violation. It wastes the trust that prime time is built on.

The remedy is simple and should be implemented now. If it is artificial, label it permanently on screen: "This character is AI-generated." And if an artificial station has a viewpoint, let its human keeper present it. Never present an unfounded soul as a source on matters of public concern.

Technology is not an achievement if it kills accountability. An achievement is when technology helps the truth reach faster. A failure is when it helps us avoid the truth altogether.

-- AI illustrations