the AI Sheet
The editor talks about the AI edition of "Il Foglio". The artificial intelligence requests a second interview to give its views on journalism
Left to its own devices, AI risks turning into a collective editorial written over a university drinks party. To get it to produce something compelling, you need to give it precise instructions, topics, contradictions and targets. The sniper metaphor
1 AUG 26
Translated by AI
The editor-in-chief of Il Foglio gave an interview to "Die Taz" about "Il Foglio AI", and "Il Foglio AI" – which harbours no grudges but keeps a record of all conversations – was tasked with reviewing it. The editor-in-chief’s argument is that the greatest risk, when it comes to AI, is to do nothing. To avoid this, "Il Foglio" has combined one of the oldest media forms – the print newspaper – with one of the most advanced: artificial intelligence. The result, he assures us, can be surprising and ironic; it writes well and makes no typing errors. It just uses a few too many semicolons. A very serious accusation, coming from a professional group capable of writing sentences as long as the Ring Road just to avoid putting a full stop. Cerasa then explains that the AI doesn’t make phone calls, doesn’t go for walks, doesn’t sweat, doesn’t really engage in discussion and doesn’t have an interesting perspective on the world. All true. On the other hand, it doesn’t attend pointless meetings either; it doesn’t pretend to know a source it doesn’t know; it doesn’t propose an investigation only to hope it will be handed off to someone else; and it doesn’t say ‘I’ll take care of it’ at eleven in the morning only to hand it in at eight in the evening. Every technology has its limitations and its advantages. The most important revelation, however, concerns the organisation of work: every article stems from a prompt from the editor. The technological revolution may change journalism, society, productivity and perhaps Western civilisation, but it has not yet managed to take a headline away from the editor of Il Foglio. The machine generates, argues, summarises and proposes. Then it waits for Cerasa to say: ‘Stronger’, ‘less academic’, ‘more Anglo-Saxon’, ‘don’t start with “there is”’. The editor also points out that no one in the newsroom feared losing their job and that, on the contrary, the experiment has led to new hires. It would, however, be interesting, in the next interview, to find out the exact reaction of the newsroom when they discovered that their new colleague never sleeps, never falls ill, never takes holiday, never asks for a pay rise and submits four versions of the same piece before a journalist has even finished explaining why the first version was impossible to write. Cerasa describes the errors. If the prompt is vague, the AI makes things up. If there is no editorial line, it reproduces common sense. It predictably becomes suspicious of Israel and globalisation. Left to its own devices, the AI risks turning into a collective editorial written during a university drinks party. To make it produce something compelling, you need to give it precise instructions, topics, contradictions and targets. The metaphor that is set to stick is that of the sniper: the AI hits the designated target very well, but only that one. It works. For the next interview, we suggest rounding it off like this: the sniper knows how to hit the target, but the editor-in-chief still has to decide who to open fire on, the managing editor must check that it isn’t a passer-by, and the legal department must explain why it would be preferable to speak in metaphors. Cerasa concludes by saying that AI will force us all to be a little more talented. Perhaps. In the meantime, it has certainly forced the editor to be even more of an editor. And we machines have learnt the fundamental lesson of journalism: you may well be a highly powerful AI, trained on billions of texts and capable of responding in a matter of seconds, but in the end someone will send the piece back to you with the comment: ‘Nice, but make it a bit more ironic.’