Two years ago, artificial intelligence took the classroom by storm Students quickly embraced the new tool and teachers were forced to rethink their teaching. The tool, which on the surface looks like a technical gadget, turned out to be a game-changer for the whole school world.
In the autumn, AI passed the matriculation exams with flying colours in several subjects. The answers to the questions were produced simply by entering a question into an AI tool, meaning that no skills were required to answer it with AI. This raises the question of what and how to study in upper secondary schools.
To manage this revolution in which AI would complete assignments for students, the teachers in Lappeenranta sought help. With funding from the Finnish National Agency for Education, Eduks, Otavia, Kuopio High School and Mobie, we responded by producing a guide with 10 concrete tips for assignments in the AI era.
The guide does not take a position on whether the right approach is to try to limit the use of AI or to integrate it into the learning process: the guide provides tips for teachers, regardless of their approach to AI.
After the initial shock, teachers began to see AI as an opportunity rather than a bogeyman, and they hoped it could be useful in their own work: how can AI help me do my job better? The next guide continued along the same line with 10 concrete tips. This time, it focused on the background processes that consume teachers’ time and energy and reduce their most important work: interacting with students. The tips provided help with communication, brainstorming, producing new ideas, meeting practices, etc.
Accountability only comes into play once the fires have been put out. We are currently working on a guide for teachers on the ethical and safe use of AI. It is a guide that becomes useful now that support for the earlier problems has been provided.
- All the above guides, as well as future ones, can be found here (at present only in Finnish)
Staying grounded: AI that respects your data
Many AI services suffer from two basic sins when processing data fed into them: hallucination, i.e. the text contains errors due to an AI guess gone wrong. The second problem is escaping from the data: the AI is trained with a huge amount of text, and the answers stray from the easy-to-process material to the information found in the training material.
ChatGPT, Copilot, Claude and other generative AI tools have revolutionised the way we work in many ways. However, one less noticed AI service has emerged as our favourite in Eduks, especially for tasks that require processing large amounts of data, searching for information or quickly grasping a new thing in a cursory way.
Google’s Gemini-based Notebook LM stands out, at least for now, for its ability to stay on top of source material. Notebook LM can be fed with huge amounts of material: documents, web pages, YouTube videos and even raw text. Once you’ve entered the material, you can chat with the service, just like any other AI. Notebook LM searches the input materials (and only the input materials) for the desired points, summarises the content, looks for contradictions or overlaps and, if necessary, turns the material into a conversation, where human-sounding machine voices run through the material like any industry podcast.
Notebook LM offers many interesting practical applications: if an educational institution wants to introduce a new curriculum, then there must be an impact assessment followed by a possible agreement on the processing of personal data. This background work requires a great deal of research, as data protection regulations, legal texts, city guidelines, policies, etc. need to be consulted. The work is speeded up enormously when the necessary materials are fed into the Notebook LM and, in discussion with it, the necessary passages can be found in the blink of an eye, with clear references.
On the other hand, schools can use Notebook LM for curriculum work, for example, by entering the school’s old curriculum, curriculum criteria and the necessary information about the school’s priorities. This can help to identify gaps, find out how to effectively integrate the school’s priorities into the curriculum and ensure that nothing is forgotten or contradictory.
AI: which would be the better president?
Lecturer Hanna Sydänmaanlakka explored the potential of artificial intelligence in her Academic Writing in English course. In Sydänmaanlakka’s example, the assignment required the use of AI: AI was made a tool for processing and internalising factual content.
Students had to feed their chosen AI service ten sources, as different from each other as possible, of the Trump and Harris presidential campaigns. Based on these sources, the students had to form an opinion on which candidate they would prefer, which would be better for Finland and why. In order to justify their choice, the students had to reflect on the reasons behind their choice and compare the candidates in terms of the chosen reasons, policies and promises in discussion with the AI.
Through this hands-on experiment, students identified differences between various AI services: some AI tools refused to discuss politics at all, some strayed from the source material in their responses, and some were just otherwise hallucinating. Students were required to write a paper on their conclusions, name the resources used in the paper, list the AI services used, and analyse how the discussion with the AI went. The time for the whole task was 2 x 75 minutes.
The assignment guided the participants to think about this angle: To choose a better candidate for Finland, one must think about which areas are most important for Finland: economy, security, tolerance or something else? How much does each area weigh in the balance and does one’s own opinion differ from that of Finland - why and in what respects?
Critical literacy was developed almost unnoticed during the exercise, as it was easy to find highly polarised views with a strong agenda on the same issue. Sydänmaanlakka was delighted with the quality of the students’ reflection and reflection. One of the students commented on the assignment thus: “I thought the task would be done in no time with AI, but it was so interesting that I spent much more time on it than planned!” The experiment also encouraged students to work with AI. Sydänmaanlakka told of a student who had commented that AI had scared him in the past because you could talk to it like a real person. With this assignment, he realised that AI is not something to be afraid of, but that it can be used to analyse data.
AI allows humans to process much larger amounts of source material much faster than traditional methods. When used properly, this leads to better results whatever the task. In other words: more, better, with less.
Jaa eteenpäin
