Mastering AI means
understanding its risks
Sociologist and expert on the risks of language models
My name is Niklas Titgemeyer and I specialise in the risks and weaknesses of language models like ChatGPT. Above all I work on disinformation from chatbots, from a socio-technical perspective. Alongside my science communication I also give workshops, awareness sessions and keynotes to help organisations understand the limitations of AI and build a critical but productive way of working with the technology.
- Risks & limitations: What can go wrong when people rely on language models like ChatGPT, and why it happens more often than most audiences assume.
- How the technology really works: The mechanics of language models explained visually, with animations, graphics and metaphors. No prior knowledge required.
- AI & society: What generative AI does to democracy, public debate and the way we think, from a sociologist's perspective.
platforms
Partners and previous roles
Focus areas of my expertise
In my studies and in my work on social media I have specialised in false information and the related risks of AI chatbots. Below you will find a list of my subject areas:
- False information in general and through chatbots
- How language models work technically
- Technical weaknesses and limitations of chatbots
- Research on hallucinations and sycophancy in chatbots
- Problems that come with outsourcing our thinking to AI
- Cyber risks from agentic AI systems
- Reasoning capabilities of language models
- Radicalisation in general and through chatbots
Podcast interview
on the risks of AI
In the podcast “WUNDERPANIK”, I talk with Katharina Mager‑Micijevic and Melanie Hölting‑Eckert about the two biggest risks of working with AI chatbots.
How often they tend to generate false information (hallucinations), in which fields and use cases this happens most, and why they are so convincing when they simply tell us what we want to hear.
The interview also asks why outsourcing our thinking to AI is a problem, and under which conditions it is in fact reasonable to trust a chatbot blindly, even if that sounds counterintuitive at first.
Here’s the episode (in German) →
Service offerings
I help people and organisations to establish a responsible way of working with artificial intelligence, and with language models in particular. Because knowing the limitations of Chatbots matters far more than adopting every new AI tool. For international audiences I offer keynotes in English, on stage or online. My other services, which include workshops and power hours on the risks and limitations of AI, are primarily aimed at companies based in Germany, since this is where I live. If you are interested in those, you can have a look at my German pages on workshops.
Free learning materials on AI
If you are an individual, or a company that would like to get a first impression of my work, I recommend taking a look at the educational material I share for free.
hallucinations, sycophancy, practical recommendations. Free video course on LLMs (German) 26 chapters and around 7 hours on how language
models work technically. No prior knowledge needed. Social media Education and science communication on AI.
Guided by four clear principles. Sources for all videos Every claim traceable and verifiable.
Here you find the searchable source references for all videos.
Take the first step.
I am happy to come to your company, your conference or your team to give a talk.
I reply to every enquiry within 1 to 2 working days.