Collaborative Learning Day "The Use of Artificial Intelligence in Research"
Researchers and research administration professionals from Tallinn University gathered on November 7th for the second Collaborative Learning Day.
Opening presentation by prof. Marek Tamm "How Artificial Intelligence Might Change a Researcher's Life"
Opening speech by professor Marek Tamm addressed the question of AI as an assistant of the researcher and/or as substitute for the researcher. Tamm also gave examples of his own experience in AI use in different research related activities like information gathering, translating, editing, reviewing and project writing.
Group work
Participant feedback indicated that artificial intelligence (AI) found applications at every stage. However, AI appeared more adept at some stages than others. The AI was seen as most beneficial in literature analysis, particularly for generating summaries of research articles and books. At the same time, it was somewhat challenging to use AI-generated text directly鈥攊t often required editing. AI's phrasing ability drew considerable criticism, as participants were not impressed by the text鈥檚 tendency to be filled with clich茅s. Some participants, however, found ways to make the generated text more specific, thereby improving its usability.
Among the more interesting examples was AI鈥檚 ability to help overcome "blank page syndrome," or writer's block. Conversations with AI allowed participants to quickly organize their thoughts or open new avenues in their thinking. Conversely, participants emphasized the need to maintain critical thinking, and trust in information outside their field of expertise was low, as there was no intuition or assurance that the generated text was accurate. Several participants reported instances where AI generated non-existent (though formally correct) citations when asked for references.
Looking to the future, questions remain regarding rules for AI use and ethical considerations. Participants expressed a desire for additional knowledge in these areas.
Panel discussion
Several of the questions raised by Marek Tamm were developed further in the panel discussion 鈥淎rtificial Intelligence in Research: Opportunities and Ethical Challenges鈥. Both Katrin Niglas (Vice-Rector for Research) and Katrin Tiidenberg (Professor of Participatory Culture) voiced her concerns about bridging and addressing the generational gap: today's researchers who have been trained without the AI will be able to understand the difference, spot ethical challenges and false information, but how to teach new generations the manuals ways of doing in order to have the comparison with working together and with the help of AI.
Andres Karjus (Lecturer of Digital Humanities and AI) stressed the importance of weighing the risks of using AI against not using it and thus being left behind. Since everybody agreed that Ai is here to stay and inevitably will influence our ways of doing research, they also stressed the importance of reflecting what researchers will do with the potentially freed time and energy. Katrin Tiidenberg pointed out that it is important that we avoid the situation where we write instead of one average article or three average articles, because at the end there will be no one to read them.
When Sonia Sousa (Associate Professor of Interaction Design) highlighted that AI is just a tool and it is up to us, humans, how we use it then Krister Kruusmaa (Guest Lecturer of Digital Humanities and AI) Krister Kruusmaa pointed out that usually tools and technology should be neutral. But LLMs (large language models) are not neutral; they have internal biases and the more they will be integrated into our everyday systems the less we will notice and be aware of the biases and this is something to be cautious about.
Andres Karjus stressed that many problems that we tend to 鈥渂lame鈥 on AI are actually questions of ethical and transparent behavior or fall under data protection matters (GDP). It was also interesting to learn that humans tend to be either overly positive (utopian) or overly dystopian towards new technologies and thus we cannot, according to Katrin Tiidenberg, trust our collective responses to technology.
For more proficient AI users, Andres Karjus recommended running language models locally rather than using cloud services when dealing with sensitive or personal data, if possible.
For creating the policies and strategies for Tallinn University Sonia Sousa stressed the importance of a vision. Katrin Niglas highlighted that it is relevant for TLU as a young and vibrant university to be at the forefront of experimenting.
According to Katrin Tiidenberg, it is also interesting how talking about AI use in research helps us to bring to light the different understandings of what are the core parts of doing research and how differently we do understand the role of each of them.