“Has Helmut Schmidt ever been mayor of Hamburg?”

According to the current version of ChatGPT, Helmut Schmidt, Germany’s former Chancellor, once has been the first mayor of Hamburg. Fact is – he never was. So, what does this teach us about ChatGPT? That was one of the questions answered in a lecture on 17 May on The Age of AI has Begun given by Christian Hille, board member of Fürstlich Castell’sche Bank, and Daniel Jung, entrepreneur and expert in tech-based learning solutions. The event in Frankfurt School’s Audimax was moderated by n-tv journalist Sabrina Marggraf and was met with great interest among students, alumni and Frankfurt School staff.

Christian Hille structured his presentation along five topics and questions:

1. Statistics on ChatGPT
2. How smart is AI?
3. Quo vadis 2023-2050?
4. What is machine learning, deep learning and AI?
5. How does ChatGPT work?

His answer to the last question: “ChatGPT is based on large language models (LLM) and on ‘prediction’. It predicts the next words, but it does not have any real knowledge nor creativity. Despite this fact it can create very useful answers and help when working and learning. You need to know how to frame your questions and also the tool’s limitations.”

Christian Hille’s presentation was followed by Daniel Jung, who talked about New learning & applications of AI. One of his key messages: “In an era of information overload, AI gets you quicker to the information you need.”

Live applications of ChatGPT covering a broad range of topics and tasks supported this message: For example, Christian Hille asked the tool to provide an overview and analysis of Nvidia’s business model – a task which ChatGPT performed remarkably well. However, in another task, it clearly failed: Asked, if Helmut Schmidt has ever been mayor of Hamburg, ChatGPT answered “yes” – which in fact is not correct.

So, the lecture’s key messages have been clear: In a nutshell, when using an AI like ChatGPT it is crucial:

1. to ask the right questions, and to frame it in the right way,
2. to be careful and check back the AI’s results,
3. and to be open to learn and not to stop thinking yourself.

The lecture was followed by a lively discussion with the audience which continued over drinks at the closing reception.