Why Construction News is using AI to assist with our writing

Technology is changing. And so is Construction News.

Our editorial team has been using artificial intelligence (AI) behind the scenes for some time now to summarise long reports and data as part of our research process, as well as produce some illustrations.

But we have not used it to produce written content. Until now.

We are launching an experiment to use AI to assist with the writing of some simple news items. The aim is to speed up the production of our least complex content, freeing up our journalists to focus on the high-standard, labour-intensive journalism that CN is renowned for.

If this experiment works, you will see an overall increase in both the quality and quantity of our content.

We will use our AI assistant, Harmsworth, to write some news stories from press releases and other notices, on simple topics, issued by trusted sources.

These will be selected in the normal way by our team as being of interest to CN readers.

We will not get AI to tackle more complex source material, such as company results or legal claims. We will not use it on long-form content such as features and opeds.

Keeping it simple has two advantages:

a) The input is easier for AI to handle, reducing the scope for it to make mistakes.
b) The output is easy for our team to check against the original source material.

Once written, AI output will be checked by human eyes, edited if necessary and published. Sometimes we will add additional human content to help the reader understand the context to the story.

We believe firmly in transparency. All of our experimental AI stories will be clearly labelled to the reader. Where possible, we will publish links to the original source material, so the reader can compare the output to the original source.

For the period of this experiment, all of our AI-assisted content will be free, and can be read without the need to register or subscribe to CN (although we recommend you do so for the really good stuff).

We would encourage you to flag up any issues via cnedit@emap.com or by commenting on individual stories to help us improve this service.