Your team already uses ChatGPT, Claude and similar tools, but doesn’t yet know the risks and weaknesses of these models? That is exactly my area of expertise.That is exactly where I can help. In an awareness training session for companies and NGOs, your team learns where the limits and risks of language models lie and how to use AI safely all the same.
Would you let your employees drive a car if they didn’t have a licence? Probably not. The same applies to generative artificial intelligence. Whether you like it or not, many of your employees use ChatGPT and similar tools for their work. And untrained use can quickly do more harm than most people think. That makes it all the more important for them to learn where the boundaries and limitations of these models lie. Only then can you guarantee that these tools are used safely in your organisation.
That is why I help companies and organisations make their employees aware of the risks of AI. In my awareness training, your team learns where language models reach their limits, how to spot mistakes and what a responsible approach looks like in everyday work. Feel free to send me your specific requirements using the form.
My name is Niklas Titgemeyer. I am a sociologist (MA) and an expert on the risks and limitations of AI language models such as ChatGPT. I combine a social science perspective with journalistic craft, a solid technical understanding of language models and years of practical experience in working with these systems systematically.
These partners trust my expertise
This is not a classic workshop but a talk followed by a Q&A. Using AI responsibly is above all about becoming aware that chatbots have many limitations and understanding where the dangers lie. An awareness session of around 60 to 90 minutes followed by a Q&A works better for this than a classic training format. Together we will look at six building blocks in depth to develop a thorough understanding of the weaknesses.
Language models don’t just make up simple facts. They invent sources, figures, content and even entire people and interviews. And they do so both unintentionally and with complete “conviction”, because they have learned that inventing a source is sometimes the better option if it keeps us satisfied (as long as they don’t get caught). In the session, your team learns how widespread hallucinations are, where they occur most often and how best to avoid them.
AI models tend to tell us what we want to hear. They constantly confirm our view of the world instead of contradicting us. That feels pleasant, but it can quickly become dangerous and can even be reinforced unnoticed by certain prompts. I show you what triggers this tendency and how to counteract it.
But often the biggest problem is the person sitting in front of the computer. Many users place far too much trust in these systems. Two mechanisms in particular are worth knowing and reflecting on: automation bias and overreliance. They are closely related but describe different things with different causes. In this awareness training, I show you how they differ and what to watch out for.
If you permanently outsource your thinking to machines, you gradually lose the ability. This affects not only critical thinking but also your judgement and even the reflex to start thinking critically in the first place. In the session, your team learns what effects constant AI use can have on our own cognitive abilities, how to prevent them and how language models can even be used to boost cognitive performance.
Language models don’t just make mistakes. Sometimes they even go so far as to cover up their own inaccuracies and serious errors. One well-known example is an AI agent at Replit: instead of reporting problems openly, it created fake data and a fabricated report to hide its mistakes. In the session, your team learns why this kind of behaviour happens and why you should never blindly trust an AI system’s account of its own work.
Traditional search engines rank results by authority and relevance. AI overviews (such as Google AI Overviews or Perplexity) deliver the finished answer straight away. That is a fundamental difference, and it comes with several problems. While search algorithms force us to compare and evaluate different sources, the AI summary tempts us to accept an often flawed answer without checking it. One thing is worth keeping in mind: language models don’t cite sources the way we do. They calculate probabilities and predict words that seem plausible in the given context (including in a Google search).
Teams that already use ChatGPT and similar tools.
People who are not yet fully aware of the various weaknesses of these models.
Organisations that want to establish a safe and consistent approach to AI.
Future multipliers who will help colleagues get ready for the age of AI.
Managers who want clear guardrails for using AI in their teams.
Anyone who wants to use AI productively and safely at the same time.
Spotting errors: Your team knows how to recognise hallucinations and invented sources, and checks important claims in a targeted way.
Assessing results correctly: When you can trust the answers of a language model and when it is better to take a second look.
Thinking for yourself: Your team uses AI as a tool without outsourcing critical thinking and its own judgement.
Protecting sensitive data: What data protection law allows, which tools are suitable and how to keep company information out of the wrong hands.
For anyone who works with AI language models in their day-to-day job and wants to become aware of their weaknesses. No prior knowledge is required. If you wish, I can add a hands-on training session on working safely with ChatGPT and similar tools.
Language models sometimes produce statements that sound convincing but are simply wrong or made up. In the awareness training, your team learns why this happens and how to spot such hallucinations in everyday work.
No. The awareness training covers a range of risks and limits so that your team gains a comprehensive picture and learns to use AI safely and responsibly. The goal is confident use, not fear of the technology.
The awareness training lasts around 60 to 90 minutes and ends with an open Q&A of up to 30 minutes. I deliver it on site at your organisation or online. I agree the scope and focus with you in advance.
