On April 3, 2025, we were again part of the Girls’ Day. 

The Girls’ Day is a nationwide event designed to showcase career and study opportunities specifically for girls. Many universities, companies and businesses across Germany participated to introduce interested pupils to science, technology and many other subjects. For your interest, this format also exists for boys as the Boy’s Day, which focuses on career paths in which the proportion of men is lower. 

The FU Berlin participated with many workshops to show which professions or fields of study may be interesting for girls. The topics mainly focused on fields in which the proportion of women is less than 40 percent. We, the Human-Centred Computing Group, participated with a workshop focusing on chatbots and virtual assistance: "Hey ChatGPT, how do I cook my favourite dish? - a look behind the scenes of virtual assistants", a format which we iteratively improved over time.

The aim of our workshop was to teach the girls how chatbots communicate, how they learn and what to look out for when using chatbots. We started by asking the girls what they thought of when they heard the words “chatbot” or “virtual assistant”. The girls were already familiar with ChatGPT, Siri and Alexa. They had all used ChatGPT for different purposes, for example as a learning aid or to help them write a song. The girls were also already aware of some of the risks associated with using chatbots, such as the fact that data is not necessarily secure.

Before diving into how chatbots actually work, we first provided a brief overview of their history. We highlighted examples such as Radio Rex, ELIZA, Siri, Alexa and ChatGPT.

Next, we started to explore our recipe chatbot. Half of the group was tasked with imagining what users might ask a chatbot when looking for cooking advice, while the other half focused on creating answers to user questions. They wrote questions and answers on different cards, and after a few minutes the groups came together to test whether their ideas could form a conversation between the bot and the user. We found that the chatbot team had relevant answers to most of the questions from the user group. Some answers even worked for multiple questions. A key takeaway was the importance of breaking down answers into smaller chunks to ensure they addressed a single question, rather than two. We also realised that the chatbot team had not included simple answers such as 'yes' or 'no'. We learned that such simple answers are sometimes necessary. In addition, the chatbot team had left certain details - such as the name of the dish - blank to make the answer more general, so that it would apply to all kinds of dishes. This proved to be a wise choice, as it made the answers more adaptable. In the end, despite having more questions than answers, the chatbot team successfully answered the majority of the user groups' questions.

After a short break, we resumed our exploration of how chatbots learn. To do this, we provided the group with partial sentences and phrases to complete with their own ideas of how the sentence might continue. We used the Menti app, which displays all the answers given and highlights the most common ones. Everyone had a lot of fun with this game, after which we explained that chatbots learn from large amounts of data, which helps them predict what users want to know. The more data and context the bots have, the better their answers will be.

Finally, we discussed the various ways in which chatbots can be used. We explained to the girls that they need to make sure that important information that the chatbot gives them is correct. They were already aware that chatbots sometimes give incorrect answers, so we briefly explained the concept of hallucinations to them and how incorrect information can have serious consequences. The girls shared their experiences with chatbots on social media platforms like Snapchat and WhatsApp. They also mentioned they are cautious about revealing too much personal information to chatbots or online in general. 

We enjoyed learning with and from the students and hope that some of them developed an interest in computer science or related fields of study. 

Workshop of Jacobe Klein with Malte Heiser and Delia Morgan