Mastering ChatGPT: Prompt-Writing Tips for Journalists

Mastering ChatGPT Prompt-Writing Tips for Journalists

Rather than being afraid of ChatGPT and other AI, savvy journalists are taking advantage. The revolutionary AI chatbot can produce the broad strokes of an idea in seconds, saving hours of writing time. Sure, it can’t fact-check, conduct interviews, provide nuanced context, or anecdotal stories that evoke the reader’s emotions, but that’s where an expert journalist excels.

But getting sophisticated output from ChatGPT is not as simple as it sounds. Like the proverbial genie in the lamp, the more carefully you express your ‘wish’, the more likely you are to have it granted. 

Here are our top AI prompt-writing tips for journalists:

Use specific keywords 

Imagine you’re on Google with a real craving for butter chicken. You type ‘butter chicken’ into the search bar, and endless results pop up – too many to sift through. ‘Indian restaurants’ gets you closer, but which is best? Finally, you try ‘Best butter chicken reviews near me’, and voila! A perfect match.

ChatGPT works in a similar fashion: vague prompts produce generic responses. For example, ‘What are Mediterranean gardens?’ will only give you a general description.

It’s far more likely to produce relevant content if you carefully select keywords based on your angle: ‘What should you plant in a Mediterranean garden?’, ‘Why are Mediterranean gardens good for bees?’ or ‘How to rewild a garden in a Mediterranean climate’. 

Whatever your prompt, remember to keep it clear and concise – overly complex language and long-winded sentences can confuse the AI.

Provide context 

What is the purpose behind your prompt? Do you want to argue a point, educate, advise or motivate? Are you writing an advertisement, an opinion piece, a how-to? Who is your target audience? 

One of ChatGPT’s most interesting features is that it excels when given context. And it’s fascinating how even slight rephrasing can change its output.

One of ChatGPT’s most interesting features is that it excels when given context. And it’s fascinating how even slight rephrasing can change its output.

For example, these two prompts will produce very different content:

  • Write a 600-word educational web article for parents, explaining why teen drivers need car insurance. Include a section on the best insurance types, with examples from [local city]. Write in the third person.
  • Write a 600-word upbeat web article for teens, highlighting the importance of car insurance. Include a bullet-point section on the best insurance types, featuring five examples of the best insurance for drivers from [local city]. Write in the second person.

Be careful – ChatGPT makes up content

You’ll need to fact-check content thoroughly, using at least two reputable sources for each fact. We prompted ChatGPT to ‘write a 150-word social media sales pitch for a new energy drink named Spark’. This is what it produced: 

Introducing Spark, the energy drink that will take your productivity to the next level! Say goodbye to sluggishness and hello to a burst of energy that will keep you going all day long. 

Spark is made with natural ingredients that are carefully selected to give you a boost without any jitters or crashes. Our blend of vitamins, minerals, and caffeine will provide you with a sustained and smooth energy lift, so you can power through your day without feeling burnt out.

With its refreshing and delicious taste, Spark is the perfect drink to enjoy before a workout, during a long day at work, or even while studying for exams. Say goodbye to boring, uninspiring drinks and switch to Spark to experience a new level of focus, concentration, and productivity.

Don’t miss out on the chance to try Spark today and feel the difference for yourself! #SparkEnergy #FeelTheSpark #PowerUp #NaturalEnergy #ProductivityBoost

To address just one of these dubious claims, our prompt did not mention that the drink contained natural ingredients. If an advertising agency used this, they’d have to back it up with hard evidence.

ChatGPT can only process queries about events prior to 2021 

We prompted, ‘Tell me why you cannot process information about events after 2021’, and this was its reply: 

As an AI language model, I am constantly learning and updating my knowledge through the data I am trained on. However, my training data only goes up until 2021, which means that any events that have occurred after that time are not included in my knowledge base.

To process information about events that have occurred after 2021, I would need to be retrained on more recent data. Until then, my responses will be limited to the information that I learned before the cut-off date of my training data.

As a journalist or content writer, don’t be afraid of ChatGPT taking over your job. It can’t. At least, not yet. It can’t set up leads for interviews, find experts for an opinion, integrate cutting-edge research into an article, make ethical decisions about which content to include, and which to omit, or make a reader cry or laugh. So, rather, consider AI as an aid which, if used properly, can optimise your workflow – in much the same way that bookkeeping software streamlines accounting. 

For more on how technology is changing the writing profession, read our blog post, ‘How to Stay Relevant as a Freelance Writer’.

Share:

Facebook
Twitter
LinkedIn

Read More...

writing in nature

Unveiling Why I Write

By Emma du Preez Last week, while walking my dogs at a local creek, I saw something that looked like a scene from the turn

Our Refund Policy

If within seven days of starting your course you are not happy on your course, we can either transfer you to a different course or provide a full refund.

If you request a refund after seven days and before 30 days we will charge a 5% administrative fee, as well as any bank fees and tutor fees already incurred.

We do not offer a refund after 30 days on the course.