The efficiency gains from AI are enormous”

Journalist: Andrea Schmits | Photographer: Markus Bertschi

Nowadays, there’s no getting around artificial intelligence (AI). Do you see this as a positive development or a bad thing?
A world without generative AI would be easier for the media industry. But this technology is here now, and it isn’t going away. What’s important is recognising how we can benefit from it. Ringier jumped on the AI bandwagon early on, equipped with appropriate AI rules and AI manifesto. This puts us in a good position to take advantage of the opportunities and reduce the risks. After all, AI is a double-edged sword.

How has AI changed the media business?
AI was and is a game changer – but it’s a gradual process. In the Blick newsroom, for example, we’ve been using AI for a long time to personalise the website. EqualVoice, our initiative to increase the visibility of women in the media, is also based on AI. A semantic algorithm counts how often women and men are portrayed in the media and analyses the context – whether stereotypes are being used, for example.

You talked about personalisation. Isn’t there a risk that users will only see content that already fits their previous views and preferences?
AI recognises which content interests the readership and weights it accordingly. This reinforces the formation of bubbles, especially on social media, and in doing so cements social factions. With us, it’s different – the most important topics are always curated by a human. This is in line with our journalistic standards and our social responsibility as a media company. Media should help to increase collective knowledge and build bridges between social groups – and not use individual relevance merely as a yardstick.

Since mid-2024, AI can also read texts out loud.
That’s right. All the texts on the Blick.ch website can be read out to you by the synthetic voice of our Chief Content Officer, Steffi Buchli. This reading function increases accessibility and combines AI with a human touch. To train the AI, Steffi Buchli spent a day in the studio and read a precisely defined script. This means the voice is equipped to handle most names, foreign terms and abbreviations.

Where else does Ringier Switzerland use AI?
AI is everywhere. Our journalists use AI tools for research, summarising texts, helping with titling or transcribing audio files. Virtually all texts come into contact with AI in some form or other. Illustrations and images are also often generated with AI. The efficiency gains are huge.

Will journalists soon be redundant?
AI is changing the long-term profile of the profession. Machines can summarise agency texts or studies, for instance. But certain journalistic activities will always be done by humans. When severe storms hit Ticino last summer, for example, it was people that went there and reported from the scene. When it comes to emotions, humans will always play a role in journalism. Nothing and nobody can assail the human spirit. That’s the true value. And we must build on this core aspect in the coming years.

How can readers tell if a text is generated by AI?
If a text has been largely generated by AI, we indicate this in the author line. We’ve brought the legendary Blick beetle back to life and renamed it ‘BliKI – the smart helper’. If AI is only used as a tool, say for translations or summaries, no labelling is needed. What matters is that every piece of content is checked by a human before being published. In this sense, the human is monitoring the machine.

Ringier has entered into an agreement with Google Cloud to develop a chatbot for the digital offering of its Swiss media brand Blick. What’s so special about this chatbot?
It’s a real challenge! The chatbot engages with users directly and suggests content that might interest them, including content from other premium media within our company or from third parties. This helps them to navigate through the masses of information. At the same time, it increases the value of the news for the individual. Bringing this offering to a reliably high level is very complex. After all, AI can also hallucinate and we’re only at the beginning of this era with generative AI. Nevertheless, hallucinations are fatal for a media company. We’ve developed very strict quality guidelines. We check the answers using a specially integrated LLM. If the answer doesn’t meet the criteria, it’s generated again.

What do you mean exactly?
AI can churn out false information or link together facts in such a way that they no longer equate to the truth. As a media company, we can’t afford to do that. We try to minimise hallucinations through human checks or using an LLM that’s responsible for quality assurance.

But people often believe such false information, or it causes them to lose trust in the media.
This is a major problem. Users are inundated with content on the Internet, some of which is entirely AI-generated. Many users lack the skills to judge which platforms they can and can’t trust. This is also very difficult as a lot of the content looks extremely professional. ‘Pink slime’ websites are a real danger. They often look like reputable news portals but are used to spread false information and manipulate entire sections of society using AI-generated content. This is poisonous for our industry.

Is there a solution for this?
These websites need to be kept in check with rapid regulatory intervention. But the democratic process behind it is too slow. Politics often lags years behind. One idea would be to agree on quality criteria worldwide and to award some kind of label to reputable media companies that commit to certain journalistic process criteria. That isn’t easy to get off the ground. But the World Association of News Publishers WAN-IFRA, of which I’m President, would be ideal for a project like this.