Over the past six months, our research fieldwork across five Swedish secondary schools has involved having long conversations with teachers who are using GenAI for their work. Here are a few things that we are regularly hearing:

#1. Only a few teachers are making frequent use of GenAI …

We are deliberately talking with teachers who have used GenAI to support their work. While there is usually a sizable minority of teachers in each school who consider themselves to be GenAI users, we are finding a surprising proportion of these only making occasional use of the technology. Common reasons include feeling that teachers lack sufficient time to play around with GenAI, and/or are only seeing marginal gains when they do use it. A lot of these teachers tell us that they intend to spend more time in the future getting to grips with GenAI – perhaps in a long vacation –  but admit that there are a lot of other priorities competing for their time.

#2. Frequent GenAI users are often using GenAI to do things that they don’t want to do.

A small number of teachers that we meet could be described as ‘power users’ – making frequent use of GenAI tools and considering themselves to be saving large amounts of time. Occasionally, these teachers are what could be described as AI-hobbyists – excited by the technology and enjoying the technical challenge and ‘fun’ of building chatbots and other AI hacks. Otherwise, these teachers are regularly delegating tasks that either they consider to be relatively unimportant, that they do not enjoy, and/or that they don’t consider themselves particularly competent at.

#3. Many other teachers feel duty bound to carry out teaching tasks for themselves

The flipside of point#2 is that a lot of other teachers strongly feel that they have to carry out teaching-related tasks themselves, and that using GenAI would compromise this. Many teachers consider tasks such as lesson planning, writing feedback or creating lesson resources to be part of their professional role and/or they feel that only they have pedagogical expertise to do a good job. 

For example, teachers in our studies tell us that planning lessons is seen as a crucial learning design process that a teacher has to ‘own’ for themselves. Grading work is seen as a high-stakes moment when a teacher is exercising their authority over the student.  Providing feedback is seen as an important relational moment, when a teacher needs to engage directly and provide guidance for each individual student. As such, these tasks remain seen as things that many teachers consider they need to do themselves.

#4. Most teachers are using GenAI to either start things going and/or finish things off.

At best, many GenAI-using teachers seem to be turning to the technology at the beginning and/or end of tasks. Some teachers find GenAI as an ideal way of ‘kickstarting’ a task – providing an initial idea or draft structure when they are feeling tired or unmotivated. This impetus then allows them to get going with the work for themselves. Conversely, GenAI is used by some as a way of finishing off or ‘polishing up’ a piece of text or resource – often improving language and phrasing. In all instances, these teachers take responsibility for carrying out the bulk of each task for themselves (for all the reasons outlined in Point#3 and Point#5).

#5. There are a lot of things that GenAI cannot do … things that teachers have to spend their time fixing

Teachers describe a range of education-related things that GenAI simply cannot do, and that require them to stop prompting and take time writing for themselves. These are things that stretch well beyond commonplace GenAI problems of hallucinations, robotic phrasing and dated training data. One key failing is lack of contextual fit, with teachers having to rework outputs in ways appropriate for their own students, classes and schools. 

For example, teachers will tell us that GenAI outputs are often pitched in ways that they know will not connect with particular students, that are framed in ways that contravene local curricula requirements, or that bump up against preferred pedagogical styles and school cultures.  Oftentimes teachers simply have a gut feeling that what GenAI is suggesting is likely to fall flat in their classroom. At best, GenAI is described as a quick way of generating basic ideas that then require the teacher to jump in and spend often considerable amounts of time and effort to rework to fit their specific contexts and needs.

#6. Most teachers are not finding GenAI to be especially exciting!

While we are deliberately seeking out teachers who have been using GenAI, only small numbers of teachers are super-enthusiastic and consider this technology to be transformative to their work. Even those who are making (semi)regular use of GenAI tend to describe it in more prosaic terms (as one teacher put it, “it’s nothing magical”). After two years, the hype around tools such as ChatGPT has died own and most teachers seen this as software that they sometimes use but can also let lapse. It is often pointed out that teachers have a lot of competing professional demands and new developments that they are having to deal with over the course of a school year. Integrating GenAI into one’s workflow is often not a high priority in comparison to dealing with these more ‘mission critical’ changes.