The lecturer, Dr Anton Gradišek, presented the development and essential characteristics of AI. Already at this stage, its usefulness was demonstrated, for example, for the preparation of timetables, the creation of thematic quizzes, stylistic variations of reading texts and summarisation. Examples from everyday life then served as a transition to the practical use of one of the AI tools, ChatGPT. We opted for its freely accessible version. Some participants ensured access to the application during the training.
The hallucination of AI
The afternoon session showed that the answers to the same questions are not always the same. We also learnt that some areas are excluded from the chatbot’s operation, such as searching and reviewing professional texts. The user is redirected to professional databases. We were surprised by the way it responds to harmful or damaging questions: it politely rejects us and tries to reason us out of them. We have, therefore, found Chat GPT helpful in streamlining our work, generating ideas and guiding us toward solutions to problems. An essential task for the user is to remain autonomous and to critically evaluate the suggestions with their own knowledge and insight before using them.
Proposed solutions may be completely inaccurate or even misleading. We tested this by asking about the name of the mayor of one of Slovenia’s municipalities. ChatGPT generates the answer word by word from the embedded data and the network connections between them, so despite a convincingly reasoned and polite answer, it gave different names. Other questions were sometimes answered illogically, which the lecturer called hallucinations. We concluded with the realisation that the use of this tool is beneficial in combination with others and not as a primary resource. Its knowledge is limited (the free version operates with data available until January 2022), and it is “very convincing in explaining an incorrect viewpoint”, as one participant succinctly concluded.
What are mentors of study circles interested in?
The mentors repeated the challenge independently. The questions focused on the content and promotion of the study circles, the animation of the different target groups, etc., and the conclusions were interestingly uniform.
Opening the door to innovation
The final evaluation highlighted two things: the need for more practical work and satisfaction with the understanding of how AI works. Participants praised the training, mainly the lecturer’s breadth of knowledge and dynamic delivery, maintaining participants’ attention and responsiveness at a high level at all times. The material, which did not contain practical examples, was perceived as more of an introduction to AI. Most of the participants have some experience with AI. It, therefore, seems to be an engaging and challenging topic, providing ample opportunity to meet diverse expectations with future possibilities for further exploration.
The organisers are delighted with the excellent response, as well as the encouragement for a new way of thinking, which requires a foundation in knowledge. We even wonder which is the better way to open the door to the innovation, ingenuity, and flexibility that today’s world demands: by gaining insight into the laws of operation (of AI or of the processes of which we are a part) or by providing practical recipes for using the tools that AI offers.
Karmen Rajar, MSc (karmen.rajar@acs.si), and Dr Nevenka Bogataj (nevenka.bogataj@acs.si), both SIAE
