AI and the media: a cautionary tale from the journalistic front-line
Journalists beware: Those quotes may not be from a human.
Any journalist who has been around for a while will tell you about the trend away from face-to-face and telephone or online interviews to written comments. Certainly in the fields I write about this has accelerated over the last couple of years.
There was a brief respite in this trend during the long months of Covid lockdowns when people were discovering the joys of video-conferencing for the first time. They were probably often desperate to see another human being, albeit on screen.
Now, we are moving rapidly towards a world where it is increasingly hard to set-up a conversation with an expert or contributor you want input from. Written comment is now almost the default response to requests, certainly in the world of specialist journalism.
I routinely send lists of dates and times for possible online interviews. Few of these are accepted as the responses come back that they will send written comments.
This leads to carefully crafted, but always cautious, often bland, comments, almost always lacking any sense of coming from the mouth of a human being. There are exceptions but they are an increasingly rare breath of fresh air.
Now, AI is inserting itself into this already depressing scenario. Rather, people are deciding to insert AI into the interaction between journalist and potential interviewee.
Written: but by whom … or what?
For one feature I wrote recently, not a single invitation to be interviewed was accepted. Instead, I received a barrage of written comments. Then, I smelt a rat.
I noticed that two of the submissions from different PR firms, representing different clients had whole paragraphs that were identical. So, I put the feature brief and questions that had been circulated via various PR websites into ChatGPT: it quickly came back with exactly the same paragraphs, largely word-for-word. Both submissions did have some additional material that wasn’t (I assume) spewed out by ChatGPT but I was faced with two sets of comments written by a robot with someone’s name appended to them to lend them a thin veneer of credibility.
I am not someone who is instinctively anti-AI. I can see benefits that will be derived from its enormous computing power. There are serious downsides. As someone who makes a living in a creative profession, of course I deeply resent the way that the tech giants have stolen huge amounts of copyright material to train their Large Language Models and believe that all governments must do more to protect and promote human creativity.
In the wrong hands the power of AI will be very destructive and disruptive but used constrictively it will also do good. We must all pause to think about the best ways to use it.
Using it to generate comments to put into the mouth of a company spokesperson is not one of those ways. Of course, PR people have done this for many years but at least that involved some human creativity. By sending me AI-generated comments they are effectively inviting someone to question whether there is really a role for a human being in that relationship in the future. That is blinkered to say the least.
As for me, I will scrutinise written submissions even harder for traces of humanity. I will also press harder for interviews when I can engage in a conversation and explore issues in greater depth, and it will be those comments that will have priority when writing up a feature.
• Picture generated by AI from the Adobe Stock collection
