Erkki Kaila
University of Turku
In early spring 2025 the head of NORDTEK AI workgroup conducted a survey to find out how university teachers use AI. In the rapidly involving situation where the AI tools in teaching and elsewhere become increasingly common, the survey shows however, that the teachers are still slow picking up and reacting to the developments. Partly the reasons can be having little (ans often justified) unconfidence in the quality of those tools, partly a healthy skepticism. However, the study also reveals that university teachers get little support themselves: universities lack or have insufficient training programs to support teachers in this new situation. Most of the respondents for NORDTEK survey are expectedly having a technical background. It is therefore even more surprising that even teachers in computed science or software engineering feel there is too little support for them to navigate the myriad of tools, regulations and norms of AI in teaching.
Demographics
The survey was based on the earlier survey¹ conducted in Finland in 2024. It was distributed via email to all universities participating in NORDTEK. The results are presented below.
A total of 126 university teachers responded to the survey. Among them, 9 (7%) were from universities of applied sciences. Nearly half of the teachers (45%) identified technology as their primary teaching area. Meanwhile, 25% listed computer science or computing, and 15% listed mathematics or natural sciences as their main areas. Thus, as anticipated, the majority of teachers were from technological fields.
The distribution of the teachers’ experience is shown in Figure 1.
As seen in the table, the majority of teachers had more than 10 years of teaching experience. About 27.8% of teachers had no experience in their field outside of the university. More than one-third had some experience (less than 5 years), and 20% had worked in the industry for over 10 years.
The survey also asked teachers about their pedagogical training. On average, teachers completed 24 credits. About 28% reported having very little training (0 to 10 credits), while nearly 17% had extensive training (60 or more credits)
Using AI in Teaching
In the second section, the teachers were asked if they had utilized AI in their educational tasks. Responses were recorded using a five-point Likert scale, with 1 indicating “Not at all” and 5 representing “Very much”. The results are displayed in Chart 1.
As seen in the chart, the usage of AI in teaching is still quite low, except for searching information. Notably, almost none of the teachers had used AI for assessing or grading students. In the open answers, many teachers mentioned that they had tried to use AI tools for generating course materials or assignments, but the quality of the generated content was not good enough to utilize in teaching.
AI Training in Universities
In the next section, the teachers were asked about the guidance they had received and whether they were familiar with the guidelines for AI usage provided by their universities. Again, the responses were recorded using a five-point Likert scale, with 1 indicating “Not at all” and 5 representing “Very much”. The results are displayed in Chart 2.
The teachers appeared to be quite familiar with their organization’s guidelines. However, it seems they received minimal training in utilizing AI. Notably, the ethical guidelines for using AI in teaching do not appear to be widely covered in their training either, which is a problem.
Knowing the Tools
In the next section, the teachers were presented with the names of some current popular AI tools and asked whether they knew them or had used them in their education. The results are in Chart 3.
As expected, ChatGPT seems to be the most familiar of the tools and also the one that’s used regularly in teaching by some teachers. Copilot has a second place in both categories. DeepSeek seems to be quite well known as well (but not much used), probably due to being in headlines recently because of the claims of its unusually light resource needs for training. Copilot is quite well known too, which is understandable because of the technical bakground of many respondents. Curiously, most of the teachers have not heard of DeepL at all.
Attitudes and Expectations
Teachers’ attitudes and expectations were observed with nine statements, which were again answered in a 5-step Likert scale (1 = totally disagree, 5 = totally agree). The statements and the averages are displayed in Chart 4.
There are some interesting observations to be made from the answers. First, it seems that the teachers mostly believe that teaching AI to students is essential but also feel that using AI increases academic misconduct. Moreover, the teachers don’t seem to be too sceptical about the future, as they hardly believe that AI will either replace them as teachers or replace humans in the field they are teaching. Finally, most teachers tend to believe that AI will change the nature of their work in the future, but on the other hand, they don’t believe that organizations know how to utilize AI yet.
Ethics of AI
In the final section, the teachers were given statements about the ethical usage of AI in education. There were a total of 13 statements, again answered in a 5-step Likert scale (1 = disagree, 5 = agree). The results are displayed in Chart 5.
First, it seems that the teachers consider privacy, data protection and ethical issues when using AI. While considering copyright is not as common, most teachers still seem to take those issues into account as well. Teachers also seem to be very cautious about reviewing the content produced by AI tools, and about being transparent about their use of AI. On the other hand, they don’t seem to believe that their students are as transparent about their usage of AI.
Conclusion
It seems that the teachers’ AI usage is not yet very common, but the attitudes and ethical perceptions seem to be adequate. Some of the key highlights about the results are listed here:
- The responding teachers had, on average, quite a lot of teaching experience and also a decent amount of pedagogical training. Most of them had at least some experience working in the industry as well.
- Aside from information searching, the use of AI tools for educational purposes is still rare among teachers. This may be partly due to the insufficient quality of the content produced by these tools.
- Low usage rates are likely due to the lack of effective training provided by universities.
- Almost everybody knows ChatGPT, and some even use it regularly for their teaching. Most are familiar with Microsoft Copilot, DeepSeek, and Google Gemini, but the other commonly used tools are not familiar to the teachers. Overall, tool usage remains very low.
- Teachers believe that AI will significantly change the nature of their work but cannot entirely replace.
- Teaching students about AI is important; however, it also raises concerns about increased cheating, highlighting the importance of instructing students on the correct way to use AI according to the rules.
- Teachers emphasize the importance of ethical considerations, security aspects, and transparency when using AI.
- They also find it important to evaluate AI-generated content before using it. Many of them reported that such material is not good enough to be used in education.
- Overall, universities should provide more comprehensive training on AI utilization, covering both the tools themselves and pedagogical methods. Given that respondents were primarily from technical areas, the need for training is likely even greater among all teachers than reflected here.
¹ Kaila, E., Lempinen, K. & Nevaranta, M. 2024. AI and Higher Education Teachers: Expectations, Attitudes, and Ethics. In FLAIEC – Finnish Learning Analytics and Education Conference, Joensuu, Finland.
