Andrew Whiteside

Opinion: Is AI good or bad in journalism?

This week the New Zealand news site NZME was criticised for using AI to write an editorial about the All Blacks rugby team. The piece was put through AI detection tools by Radio New Zealand’s Mediawatch and showed positive results for AI generated content.

This story raises some interesting issues around journalism and the profound effects AI will have on this industry and indeed society.  

One of the biggest fears around AI is that it will and indeed has led to job losses. A number of media outlets around the world have sacked writing staff and are using remaining workers to generate news stories and content with AI.

I have mixed views about all of this. My personal view is that AI is inevitably going to impact greatly on every aspect of human life both negatively and positively. I believe I should embrace AI and learn as much as I can about it while still holding onto the values and ethical standards I try to bring to my work. Just sitting back and letting the coming AI wave wash over me I believe will be a bad thing. 

Let me explain. 

As I typed the paragraph the predictive text feature of the word processing software guessed many of the words I actually went on to use. That has been around for some time and it’s a tool that is designed to make typing easier.

The thing is though, the words it chose, which I also was intending to type, were quite cliched. The types of things we all say as they are quite simple and easy to remember but not very creative.  

So I went back and retyped some of it using different words which was a more creative way of saying the same thing. In this case then, an innovative technology has in a way forced me to be more inventive. 

AI for me is a useful tool. I use chatGPT for example to do initial research on topics in which I have little knowledge and to summarise issues. But like any source of information I cross reference with other sources and check it for errors. 

Recently I also used AI to improve the sound on an interview I recorded which had a lot of background noise and echo. To me that is no different than using audio software such as Pro Tools to ‘sweeten’ audio. Likewise I am experimenting with AI generated graphics and pictures. 

Since art is not my forte and I don’t have the resources to hire artists or graphic designers AI offers me a cost effective way of creating content. 

Of course I am well aware that the impact of this is that people are losing their jobs, but I think that ultimately new opportunities will arise from the AI wave.

What does worry me is misinformation. Already AI is being used to create deepfake imagery and as the technology improves, the ability to mimic real people and put out false propaganda will cause real problems.

But here’s where things get interesting. As the NZME issue shows, there are already tools that can detect when things are AI generated, and those tools will also improve.

Having read that article, I decided to run an experiment of my own using the ZeroGPT AI checker. I chose an opinion piece I wrote about social media in February 2024 called Is My Phone Out to Get Me? and ran it through the check system. At the same time, I got chatGPT to create an op-ed on the same topic and put that through ZeroGPT as well.

The results were:

My opinion piece was judged as 9.43% AI and concluded ‘Your text Is Human written’.

ChatGPT’s op-ed was judged as 97.03% AI and concluded ‘Your text is AI/GPT generated’.

The checker picked four sentences in the piece I wrote that it judged were ‘probably’ created by AI, even though they weren’t. 

I wasn’t too upset by that as I know from research that the checkers can have inaccuracies, and also GPTs base their creations by scooping up material that is written by real human beings. Perhaps the four sentences were a little too bland and therefore judged as being written by a computer programme.

Maybe AI checkers can be a useful tool for improving one’s writing. After all, if I run anything I write through a checker and it finds sentences that could be AI, perhaps it will encourage me to rewrite them in a less generic way.

So, I am going to run this op-ed through the checker after I finish it. 

My ultimate conclusion is that AI is a fascinating tool that can only be judged as good or evil or something in between by how we humans use it. After all, the only reason bad things happen in the world is because humans make choices that affect other people. AI is a tool that we can choose to use to make our lives better and I hope we do!

And now the results of the ZeroGPT test on this piece:  0% AI GPT – ‘Your text is human written’.

Whew!

Notes: The image for this story was AI generated using Dalle.3. Subsequently to checking this op-ed on ZeroGPT I made some changes to vocabulary and minor sentence reconstructions, but all by me, a Human!

Exit mobile version