Digital Innovation Digital Publishing Top Stories
4 mins read

Automation vs. AI: What publishers need to know

Freeing up time with automation will deliver much better ROI than trying to chase the generative AI genie, even if it is well and truly out of the bottle

I’ve just finished writing a report on AI in publishing and I have to admit, I was deeply confused when I started writing it. Is AI the same thing as automation? Is automation a subset of AI? What even is AI?

Here’s the thing. After talking to people much smarter about AI than I am to write the practical AI report, I’ve realized it doesn’t actually matter, at least the automation versus AI part. The important part is that these technologies bring value to what you do as a publisher.

One of the biggest reasons for the confusion that surrounds AI is that it is not one technology.

In an excellent AI myths piece for The Washington Post, consumer tech reporter Tatum Hunter flags the fact that AI isn’t one big thing but a ‘collection of different technologies’. Charlie Becket at the LSE’s Journalism AI project described it to me as a ‘bundle of technologies’.

What ties this disparate collection of technologies together is their ability to process large volumes of information, to a greater or lesser degree, on their own. At the practical AI end of the spectrum that’s algorithms that follow strict rules to filter, extract and format information from huge datasets. The secret is a well-structured spreadsheet.

At the sci-fi end of the AI spectrum, you’re probably talking about machine learning. And that’s where the popular perception of AI sits right now, headed into the 2nd quarter of 2023, with image generators firing out pictures of the Pope in a Balenciagia coat.

Indistinguishable from magic

Generative AI appears to be sufficiently advanced to invoke Arthur C. Clarke’s third law that ‘any sufficiently advanced technology is indistinguishable from magic’. Of course, we know that being indistinguishable from magic does not actually make something magic. We know that there are all sorts of rules and algorithms working behind the scenes. We just choose to forget.

People see the output of image generation technology like Midjourney and wonder what it can’t do. But the hype has fueled the popular misconception that all AI is generative AI. Ask the machine to create something and hey presto, there it is, from the Pope in a big coat to a Ph.D. thesis.

Against that backdrop, a word like automation seems far too pedestrian. It conjures images of assembly lines or washing machines… industrial, repetitive, process driven. If generative AI demands a spell book, automation needs a flat-pack furniture manual.

But here’s the thing, automation AI has proven to be far more useful for publishers at this point than generative AI. There are an increasing number of experiments using generative AI in the newsroom. All are interesting, some for completely the wrong reasons; CNET was one of the first to announce the AI authorship of a series of almost 80 articles, 40 of which contained serious inaccuracies.

Contrast that with the robot journalists used in newsrooms around the world to create automated content day in, day out without mishap.

  • PA Media’s RADAR newswire has generated more than 600,000 articles since it was launched in 2018, each a local story extracted from a national dataset.
  • Stavanger Aftenblad in Norway covers 10,000 junior league football matches a year using automated text and a single reporter.
  • US newspaper group McClatchy provides details of hundreds of neighborhood property deals across several of its markets using just bots.

Where AI is generally thought to be software that is designed to simulate human thinking, these applications rely on software that follows pre-programmed rules to create highly formulaic content. And while it might not be as advanced as ChatGPT, it works.

Since 2021, Bloomberg has been using automation to help cover press releases and personnel announcements, using AI for the first draft of these stories. The time saved sees these types of stories published more quickly, but also frees up time for reporters to spend on more complex stories.

Commenting on Bloomberg’s use of the technology, cofounder of Aging Media, John Yedinak, said on Twitter: “By freeing up their time using automation, it allows edit teams to get on the phone, dig deeper into trends, and frankly work on the hard stuff that people really want.

Going further than that, Jens Pettersson at Swedish newspaper group NTM Media told me that robots can automate story alerts for reporters, spotlighting trends and exceptions in property sales that reporters can follow up on.

There is no doubt that AI will have an impact on how publishers work in the future. But right now, I’m staying away from the kind of AI trickery that should scare anyone that worries about misinformation. This has already seen over a thousand tech leaders send an open letter calling for ‘all AI labs to immediately pause for at least 6 months‘. If Elon Musk is calling for a temporary halt to new generative AI releases, government regulation can’t be that far behind.

Instead, I’m advocating for steady, reliable automated content bots that complement editorial efforts with stories reported from data at scale. Freeing up time with automation will deliver much better ROI than trying to chase the generative AI genie, even if it is well and truly out of the bottle.