{"id":70176,"date":"2026-04-08T08:04:00","date_gmt":"2026-04-08T06:04:00","guid":{"rendered":"https:\/\/www.mashup-communications.de\/2026\/04\/ai-vs-the-human-brain-how-to-succeed-at-storytelling-with-and-without-llms\/"},"modified":"2026-05-09T11:38:12","modified_gmt":"2026-05-09T09:38:12","slug":"ai-vs-the-human-brain-how-to-succeed-at-storytelling-with-and-without-llms","status":"publish","type":"post","link":"https:\/\/www.mashup-communications.de\/en\/2026\/04\/ai-vs-the-human-brain-how-to-succeed-at-storytelling-with-and-without-llms\/","title":{"rendered":"AI vs. the Human Brain: How to Succeed at Storytelling With and Without LLMs"},"content":{"rendered":"\n

A year ago, I announced a blog post on the topic \u201cHow do I write with ChatGPT?\u201d Back then, working with text using large language models (LLMs) was still more like a guessing game<\/a>. A lot has changed since then: whether it\u2019s researching, outlining, editing, or translating\u2014just a year later, it\u2019s hard to imagine everyday communication without ChatGPT, Claude, or Gemini. <\/p>\n\n

But as this has become the norm, a new question has arisen: What would I be better off writing without AI?<\/p>\n\n

After all, the temptation to trade creativity for short-term efficiency is more alluring than ever. Yet this is precisely where it will be decided whether storytelling will simply become faster\u2014or truly better\u2014in the future. <\/p>\n\n

\"Image<\/figure>\n\n

Quelle: Mashup Communications \/ KI-generated<\/p>\n\n

From Idea to Story: When Should I Use AI\u2014and When Should I Not?<\/h2>\n\n

I\u2019ve been able to incorporate a few of the tips and tricks I had in mind a year ago\u2014following my \u201cCopywriting with ChatGPT\u201d course at the School for Communication & Management<\/a>\u2014into my daily work. From research to polishing\u2014what role should LLMs play in copywriting today, and which ones are best avoided? <\/p>\n\n

\ud83d\udfe1 Research: A Good Start<\/h2>\n\n

How do I use AI for research?<\/strong><\/em><\/p>\n\n

ChatGPT, Claude, and Gemini provide an initial starting point and help you quickly get an overview of what has already been said on the topic and which arguments are most prevalent. But the research must go further from there. I verify sources, review the original studies, actively seek out opposing viewpoints, and incorporate additional perspectives. After all, an AI search query is no substitute for thorough, independent research. <\/p>\n\n

\n
Learn more here about AI confirmation bias and what you can do about it<\/a><\/div>\n<\/div>\n\n

<\/p>\n\n

\ud83d\udd34 Thinking: AI has no place here<\/h2>\n\n

When should I avoid using AI for writing?<\/strong><\/em><\/p>\n\n

When it comes to the actual idea, I deliberately avoid using AI. Because writing is thinking. The moment I refine a story or craft a teaser, the core of the story takes shape. That\u2019s when it\u2019s decided whether a topic is original or just generic drivel. <\/p>\n\n

\n

\u201cEspecially when interacting with journalists, originality is essential for sparking interest. After all, newsrooms now receive AI-generated story ideas on a daily basis. So if you want to stand out, you need a genuine human perspective.\u201d <\/p>\n\n\n\n

\u2013 says Alexandra Reinig, expert in PR and data storytelling.<\/pre>\n<\/blockquote>\n\n
<\/p>\n\n
\ud83d\udfe2 Text Structure: Support with Responsibility<\/h2>\n\n
Why should I use AI for text editing?<\/strong><\/em><\/p>\n\n
Working with LLMs has helped me overcome my fear of the blank page. AI is excellent at suggesting structures or phrasing. That\u2019s because ChatGPT, Claude, or Gemini don\u2019t \u201cthink\u201d\u2014they calculate which sentences are most likely to be understood.  <\/p>\n\n
However, for the results to be convincing, the target audience, communication objective, and text type must be clearly defined. That said, good prompting is no substitute for a critical review of what is said. Just because something sounds good doesn\u2019t mean it has substance.  <\/p>\n\n
\ud83d\udd34 Expressing an opinion: Taking a stand involves taking a risk<\/h2>\n\n
Can I use AI for thought leadership?<\/strong><\/em><\/p>\n\n
When it comes to opinions, I don\u2019t use AI. The same goes for headlines, key messages, or quotes: meaningful statements take a risk. They can sometimes be uncomfortable and\u2014unlike the output of chatbots\u2014aren\u2019t designed to please everyone.  <\/p>\n\n
However, thought leadership\u2014whether in trade publications or on LinkedIn\u2014only works when you take a genuine stance.  <\/p>\n\n
If you want to be seen as an expert, you have to take a stand. No machine can take on that responsibility.<\/p>\n\n
\ud83d\udfe2 Storytelling: LLMs als Werkzeug f\u00fcr Dramaturgie<\/h2>\n\n
How do I use AI for storytelling?<\/strong><\/em><\/p>\n\n
When it comes to actual storytelling, AI can be helpful if used thoughtfully. This is where prompting comes into play again. For example, I can ask the tool to put itself in the shoes of a scene or a specific problem situation. ChatGPT, Claude, or Gemini can take on a specific role when formulating text, such as that of Goethe (rather useless) or that of editor XY at the trade publication XY (very useful). Tone and perspective can also be specified\u2014for instance, analytical, casual, or phrased in a way that even readers outside the field (\u201cWrite it so your grandma could understand\u201d) can follow.    <\/p>\n\n
\"-\"<\/figure>\n\n
Source: Mashup Communications<\/p>\n\n
However, this only works if I have a firm grasp of the craft myself. If you don\u2019t understand narrative structure, shifts in perspective, or how to address your target audience, you won\u2019t be able to craft a compelling story\u2014even with the help of AI. <\/p>\n\n
\ud83d\udfe2 Fine-tuning: Efficient, but no substitute for the dual-control principle<\/h2>\n\n
How do I use AI for proofreading?<\/strong><\/em><\/p>\n\n
My most common prompt is probably: \u201cImprove.\u201d AI is very useful for proofreading. It can quickly check grammar, clarity, and stylistic polish. However, it does not replace the classic dual-review process. A machine has limited ability to recognize strategic nuances, sensitive phrasing, or implied meanings.    <\/p>\n\n
\ud83d\udd34 Images: Authenticity Remains Human<\/h2>\n\n
How can I use AI for visual storytelling?<\/strong><\/em><\/p>\n\n
Have I ever used AI-generated images? Of course (for this post, for example). Does working with AI open up possibilities for surprising visual storytelling? Definitely, as long as the creative impulse comes from us. Does AI replace authentic imagery?     No! Visual storytelling still requires real people and real-life situations that we can relate to. Those who rely exclusively on AI for the sake of efficiency or to circumvent copyright laws risk having their audience quickly grow tired of it and turn away. <\/p>\n\n
\ud83d\udfe2 Translations: Fast, but subject to review<\/h2>\n\n
Which AI tools are useful for text editing?<\/strong><\/em><\/p>\n\n
When it comes to translation, AI is a helpful tool, especially for initial drafts. Nevertheless, I carefully review the results. Cultural nuances, industry-specific terms, and sensitive contexts require a human touch.  <\/p>\n\n
It is also advisable not to rely solely on a single tool. Depending on the specific use case, specialized services such as DeepL can provide more accurate results when it comes to translation. A deliberate mix of tools improves text quality.  <\/p>\n\n
Reflection<\/h2>\n\n
Looking back on the training I completed a year ago, what I\u2019ve taken away from it is a sense of calm. Yes, AI is changing our profession. But it doesn\u2019t replace what makes for good communication: good judgment, a feel for language, and curiosity.  <\/p>\n\n
The key question, then, is: Where is AI helpful \u2014and where does it undermine my voice?<\/p>\n\n
The new pace is tempting. But relevance doesn\u2019t come from speed; it comes from taking a bold stance. Those who read widely, write frequently, and engage with different perspectives will also be able to evaluate machine-generated suggestions. It is helpful to specifically ask the AI for criticism or to have its responses evaluated according to journalistic standards. However, using the chatbot professionally also involves continuously examining the underlying training data and language patterns to better recognize the AI\u2019s strengths, weaknesses, and limitations.    <\/p>\n\n
Conclusion: AI is a tool\u2014not a voice<\/h2>\n\n
AI tools like ChatGPT, Claude, or Gemini speed up the writing process, but not the thinking process. Large language models (LLMs) analyze linguistic patterns\u2014they do not develop their own perspectives or take risks. Yet that is precisely what is crucial for originality, thought leadership, and journalistic relevance.  <\/p>\n\n
The focus is therefore shifting: it is not writing itself that is becoming the core skill, but rather thinking, analyzing, and evaluating.<\/p>\n\n
Those who use AI strategically,<\/p>\n\n