verdict
So how well did the AI reporter do its job?
The most basic aspect of good reporting is being accurate and fair. In that regard, AI Reporter scored pretty well. Most of its story ideas were reasonable and factually correct, with no obvious evidence of bias.
Bias: AI may exhibit biases based on the data it is trained on, and may reflect biases that already exist in society.
What impressed AI Reporter the most was the TV script. He demonstrated an outstanding ability to write short, punchy texts about climate change and its health impacts, and also suggested the right kind of video to accompany it.
But things can go wildly wrong, at least during the pitch phase.
Another image created by AI for the milk story with stable diffusion
Another image created by AI for the milk story with stable diffusion
For example, the most concerning mistake in the milk and road safety story was that it was made with confidence and conviction supported by a plausible but wholly fictitious account.
It suggests that AI is “seeing hallucinations”.
hallucinations: This is when AI produces incorrect or nonsensical information, often because it doesn’t truly understand the data it’s given.
So was that journalism the sort of thing a human reporter would produce for a major news outlet like Sky?
From that perspective, I wouldn’t worry too much about the AI we’ve developed taking my job. The story wasn’t wrong, but it wasn’t very “new” or imaginative. The “angle” you chose was reasonable but not original.
This is no surprise. AI that predicts sentences has no “imagination”, and the only “news” that can be given is what Sky and others have already announced, and it is not expected to make inferences based on a broader understanding of society.
However, we used “off-the-shelf” components to build an AI reporter in a matter of weeks. With more time and a team of AI experts, it will be possible to further hone the skills of the AI Reporter.
As Professor Beckett of LSE’s JournalismAI says, ultimately human resources often waste journalists’ time by performing more mundane production tasks or helping with emails and drafting scripts. of reporters may be able to spend more time on duty.
“Savings from efficiencies could be spent sending more reporters to interview people, or writing more imaginative, empathetic, or more opinionated stories. Machines can’t do that,” he says.
As for the AI ”face”, well, it’s not without fault. Clips had to be redone at times because Hannah’s accent could be all over the place. Sometimes authentic, sometimes a little Irish, sometimes American. But the possibilities clearly exist, as we asked for clarification.
There is no doubt that journalists will have to adapt to AI in new and exciting ways. We are just beginning to realize that it can disrupt my profession, and countless others. What struck me was that even half a year ago we wouldn’t have had the AI tools to produce anything like what we did here. Who knows what we have to do before Christmas.
But for now, I can breathe a sigh of relief, put my salmon fishing rod back in the cupboard, and know that my job is stable for now.
