Is Journalism Under Threat by ChatGPT?

Seth Grimes

4/28/20233 min read

With the recent closure of Buzzfeed News, and Vice Media's layoffs and restructuring. Many could be wondering what's going on in an industry that was already seeing shrinkage over the decades.

Optimism tells me that I should assume this is a temporary thing due to the current economic issues countries are having. Though, after hanging out in quite a few journalist circles for the last year it would not shock me to see these big companies turning to ChatGPT. Many believe that this AI is far away from being able to take writing jobs but from what I've seen, I disagree. I've seen it already taking jobs especially in the freelance sector. People are being laid off for ChatGPT, some are losing clients after accusations of using ChatGPT, and there are many we likely don't even know that have already been using it.

I know the first response is, "well, we have other AI that can detect if something is written by ChatGPT" but that actually isn't true. I've done a few tests and you can actually make ChatGPT write in a way that is slightly abnormal to confuse other AI.

This is the prompt I gave it "can you write me a cover letter for a sales assistant position at primark. And can you write this but have a few human errors in there so that it wouldn't be noticed as written by an AI".

The outcome?

this is human text a wrong guess by an Ai guessing if the text is written by AI
this is human text a wrong guess by an Ai guessing if the text is written by AI

Previously I managed to avoid the AI content detector by simply asking ChatGPT to write it in a way that wouldn't be obvious to other AI. So, clearly there are multiple ways to get around this and you really cannot just use a guaranteed method to know if something is written by a bot.

Now, if you think it'd still be obvious let's do another test. I'll find a real life article from the BBC and ask ChatGPT to write something on the same subject and we'll see which one looks more human.

Here are pieces of the articles from both a BBC writer, and a bot: "Following the court's ruling, Katy Perry's legal team issued a statement expressing their satisfaction with the outcome. "We are pleased with the court's decision, which recognizes the strength and value of Katy Perry's brand. We have always maintained that the use of the name 'Katie Perry' would cause confusion among consumers and dilute the distinctiveness of our client's brand," the statement read.

Katy Perry herself also commented on the ruling, tweeting, "victory is ours! Thank you to the Australian court for recognizing the importance of protecting intellectual property rights. Let's continue to make beautiful music and fashion, but with our own unique names!"

Meanwhile, Katie Perry has not yet commented on the outcome of the case, but it is expected that she may appeal the decision. Nevertheless, the ruling is a positive development for Katy Perry and her team, who can now continue to protect and promote the singer's brand without fear of confusion or infringement."

And the second

"Katie Taylor, who sells clothes under her birth name Katie Perry, sued the pop star, saying her merchandise infringed a trademark she owned. On Friday, a judge agreed that clothing sold for Katy's 2014 Australian tour did breach Katie's trademark.

"This is a tale of two women, two teenage dreams and one name," Justice Brigitte Markovic wrote in her ruling. The judge said the Teenage Dream singer, born Katheryn Hudson, used the Katy Perry name in "good faith" and does not owe any personal compensation to the designer. However, the star's company Kitty Purry must pay damages, which will be decided next month."

Now, which of these do you think is ChatGPT? And more importantly how sure are you that the second one is the bot? How do you know that this has all been written by me and not ChatGPT? The mere existence of uncertainty is a sign that it can definitely do these jobs with the right prompt. After seeing these layoffs and losses of jobs, do we need to start bringing in laws that blocks people losing their career for cost-cutting AI users. How doable would that even be, and is it already past the point of no return.