Split Earlier this year, I was fired and replaced by a robot, and the managers who made the decision didn’t tell me or anyone else affected by the change that it was happening.

The job I lost began as a happy and profitable relationship with Cosmos magazine – Australia’s rough equivalent of New Scientist. I wrote occasional features and a column that appeared every three weeks in the online edition.

Everyone seemed happy with the agreement: my editors, the readers and myself. We had found a rhythm that I was convinced would continue for many years to come.

That wasn’t the case. In February, just days after I submitted a column, I and all the other freelancers at Cosmos received an email telling us that no more submissions would be accepted.

It’s a rare business that can profitably serve both science and the public, and Cosmos was no exception: as far as I know, it was kept afloat with financial support. When that funding ended, Cosmos ran into trouble.

I accepted the economic realities of our times, mourned the loss of a great opportunity for my more scientific investigations, and moved on with my life.

However, it turns out that wasn’t the whole story. Six months later, on August 8, a friend texted me with news from the Australian Broadcasting Corporation. In summary (courtesy of ABC):

Cosmos had been caught using generative AI to write articles for its website – using a grant from a non-profit that runs Australia’s most prestigious journalism awards. That’s why my work – writing articles for that website – had disappeared so suddenly.

But that’s not even half of it. The AI ​​was most likely “fed” my articles – via the “Common Crawl”, the giant tarball containing almost everything ever published on the web – to ensure the accuracy of this content.

I wasn’t just fired and replaced by a robot. This robot was programmed to be my replacement.

The article goes on to report that Cosmos’ editors-in-chief were unaware of this. It all happened quietly – which speaks volumes about how this suggestion would have been received had it been communicated to the staff responsible for working with freelancers. Cosmos’ mea culpa on the incident laments the lack of communication ahead of the work that led to AI-written articles appearing.

What an understatement.

Editors know that audiences want to read texts (like this one) written by humans. The boring “middle” content generated by an AI may be good for a summary, but it lacks the human touch. It’s enough in a pinch, but not particularly satisfying.

Cosmos has decided to produce the crap that fills every marketing channel on the web, because generative artificial intelligence delivers more of what marketers want to show us – but little of what people want to read.

Cosmos had the courage to label AI-generated articles – more transparency than we expect from other publications that operate in secret and become one-man shows with a single person managing the production of a massive content farm.

Wiley closes 19 scientific journals due to problems with AI paper mill

READ MORE

There are techniques to watermark such AI-generated content – readers could easily be warned. But this idea has already been rejected by OpenAI CEO Sam Altman, who recently stated that AI watermarking threatens at least 30 percent of the ChatGPT maker’s business. Companies don’t want to admit that they produce crap and spam us with it.

In the absence of this kind of detection, what we need is something more like a chain of origin that shows the path of these words from my keyboard to your eyes, revealing the process of writing, editing and publishing. With this kind of transparency, we can recognize the human factor.

This human touch has never been unrivalled. Now that it is here, it has immediately become the most valuable thing a reader can experience. That should be reason enough to make it a reality. ®