Izvleček
V prispevku je predstavljena učna ura, izvedena v okviru oddelčne skupnosti sedmošolcev in knjižničnih informacijskih znanj. Izvedena je bila z namenom, da se učencem s konkretnimi primeri pokaže kritično rabo umetne inteligence. Razredničarka je učencem predstavila možnosti uporabe le-te, s poudarkom na njenih prednostih in slabostih. S knjižničarko sta po uvodnih informacijah učence povabili k praktičnem delu z vnaprej pripravljenimi odgovori ChatGPT-ja na vprašanja o Francetu Prešernu in njegovem življenju. Odgovore umetne inteligence (UI) so primerjali s pripravljenimi knjižnimi viri. Po kritičnem pregledu odgovorov UI in njihovem primerjanju z ustreznimi tiskanimi viri so učenci sami ugotovili, da je ChatGPT nezadosten vir podatkov in da je treba dobljene odgovore natančno preveriti in zahtevati vire uporabljenih informacij. Z opravljenima učnima urama so učenci sami dokazali, da je odgovore UI vedno treba kritično presoditi.
Abstract
Workshop on Critical Evaluation of Artificial Intelligence Responses Compared to Published Resources
The article presents a class organized for seventh-grade students on the topic of library information knowledge. It was implemented with the purpose of demonstrating to the students a critical use of artificial intelligence on concrete examples. The class teacher introduced the use of AI to the students, emphasizing its advantages and disadvantages. After giving the initial information, she and the librarian invited them to practical work using the ChatGPT answers, prepared in advance, to the questions about France Prešeren and his life. Artificial intelligence (AI) answers were compared to the prepared library resources. After a critical evaluation of AI responses and their comparison with applicable published resources, the students themselves concluded that ChatGPT is an inadequate source of information and that AI responses must be thoroughly verified, requesting also the sources of the information used. By participating in the workshops, the students were able to prove themselves that AI responses must always be critically evaluated.