Mobile navigation

COLUMN 

Beware Copilot

Copilot and similar AI drafting tools are amazing, but journalists should be extremely wary of them…

By James Evelegh

Beware Copilot

It could ruin your career.

Imagine the scenario. You’re a time-poor book critic with a commission to review Moscow X by David McCloskey (a 2023 book, which I happen to be reading at the moment, very good by the way). Your editor needs it by tomorrow and you haven’t even read it, let alone put pen to paper. It’s 400+ pages long. It’s not going to happen.

You open a new document in Word, with that invitation at the top which you’ve routinely ignored up until now, ‘Describe what you’d like to draft in Copilot’.

What the hell, you think, let’s see what happens.

You type out: ‘Write me a 300 word review of the book, ‘Moscow X’’ and hit return, not expecting much, but you never know.

Less than a minute later, 50 seconds to be precise, bingo! A ready-to-go review.

It’s well written – an excellent summary of the plot line and characters: “… modern spy thriller that feels uncomfortably plausible… The novel ranges from the polished legal and financial ecosystem that launders oligarch wealth to the darker, violence-laced edges of Russian power… Stylish, gritty, and intellectually alert … Recommended for fans of le Carré-style realism who still want a contemporary, high-stakes page-turner…”

What to do? You need the commission, you don’t want to let the editor down, and the review is really good.

Desperate measures. You send it in, it gets published, you get paid, phew, no harm done, move on.

Except, except, you’re left with a nagging doubt. You feel uneasy, but are not quite sure why.

Then, everything starts to unravel. A reader points out that the review draws heavily from the blurb on the back of the book and the published testimonials on the author’s website. Worse still, a fellow journalist who reviewed the book for another paper has called you out for plagiarism, saying that whole paragraphs in your review are strikingly similar to theirs. He has raised it with your editor, who is none too pleased.

The editor apologises, adds a footnote to the bottom of your piece crediting and linking to the original review, and releases a statement saying you won’t be writing for them again.

You’ve always shown a healthy disdain for plagiarists in the past, but that is what you’ve become.

Not good.

That entirely apocryphal story has echoes of a real story this week of a reviewer for the New York Times being called out by an eagle-eyed reader, who pointed out that their book review was very similar to a review carried by the Guardian. The NYT quizzed the author, who owned up. He said he was “hugely embarrassed” and had “made a serious mistake”. The NYT has since added a footnote crediting and linking to the Guardian piece and stated that the author will not be writing for them again.

The lesson? If it’s got your byline, make sure it’s written by you, not by Copilot.


You can catch James Evelegh’s regular column in the InPubWeekly newsletter, which you can register to receive here.