‘Help!? My students created an evil AI’: on the irony of speculative methods and design fiction

ABSTRACT

This article contributes to moving forward current discourses about speculative methods by explaining how they can become critical in practice. It explores how the collaborative construction of speculative technological solutions (in this case relating to AI in education) can reveal an implicit acceptance of restricting imaginaries. In three design fiction workshops, a total of 36 teacher students were given the task to create an AI-powered technical solution to a self-identified and urgent problem in education. These design workshops revealed several interesting ‘ironies’ relevant to both actual technical development as well as speculative methods about technological futures. The ironies include an ‘ironic imitation’ of the discourses and norms that are perceived as surrounding contemporary technology. These workshops further reveal and question discourses about desirable learning, and as such these ironies are productive in illuminating power asymmetries, thereby creating space for important failures, resistances, and disruptions.

KEYWORDS:

• Speculative methods
• design fiction
• education futures
• AIEd
• anticipated futures

Introduction

Schools grapple with the concerns and promises of artificial intelligence and autonomous technology, driven by visions of calculable futures (Williamson, Macgilchrist, and Potter 2023). As the anticipated AI future accelerates and becomes more uncertain, there is political (Kim 2024) and economic will to engage in the work of futuring (Komljenovic et al. 2023). To navigate this landscape, methods like speculative fiction, critical education futures, and design fiction are gaining traction (Bayne 2023; Cerratto Pargman, Lindberg, and Buch 2022; Hrastinski 2023; Linderoth, Hultén, and Stenliden 2024; Suoranta and Teräs 2023; Teräs, Teräs, and Suoranta 2023). Recognising education’s role in shaping futures (Ross 2023), this article explores how AI may emerge from discussions about desirable education, potentially leading to both individualising and oppressive AI in education (AIEd). Speculative methods serve as tools not only for envisioning positive futures but also for exploring evil, weird, and failed simulations, thereby strengthening a critical understanding of AIEd (Costello et al. 2024). These methods can question the equity effects of edtech by highlighting its semiotic and structural foundations.

Researchers have previously highlighted the role that language, meaning-making, myths and metaphors have had for guiding the societal impact of AI (e.g., Bareis and Katzenbach 2022; Natale 2021). Further, Knox et al. find that ‘a narrative of ‘personalisation’ often appears to drive AI research and development in education’ (Knox, Wang, and Gallagher 2019, 4), which ‘also appears to align rather seamlessly with contemporary views of ‘learner-centered’ education’. Macgilchrist et al. (2024) underscore that positioning edtech as a tailored solution to a particular problem is not just a practical matter; it is deeply rooted in politically charged histories and futures marked by exclusion and oppression. Additionally, Macgilchrist (2018, 84) identifies the enactment of a ‘cruel optimism’ when data-driven systems are coupled with pledges of equity. She writes:

The data stories point to the constitutive paradoxes of education which are ‘sticky’ and cannot be cleaned away. The stories of generating data, protecting data and using data show hope for change. They also show how the fantasy of equality is projected onto a socio-technical mediator (a personalised literacy platform, data privacy practices, an active parent armed with data visualisations) which enables a small interruption in inequality, but also blocks attention to the weakening of the fantasy of an equitable life in today’s increasingly post-democratic world.

Similar confidence that AIEd can solve perennially difficult educational questions is found in the study reported in this paper. However, what begins as a technological utopia to create the very best conditions for learning can, as the results show, quickly create an idea of increased requirements and new necessities to achieve specific goals, regardless of the costs involved.

In terms of pedagogical benefits, I identify the process of constructing technological, albeit fictional, solutions as an important part of making visible the socio-material effects of emerging technologies. Instead of abstract reasoning about potential drawbacks of technologies, which often might remain aloof and remote, this process makes the latent power asymmetries embedded in technology more tangible. The speculative technical solutions also embody certain values and thoughts that can be collectively criticised and reflected upon. As such, my research questions can be found on three levels relating to (1) problems, (2) solutions, and (3) critical understandings of these socio-material problem-solution configurations. They are formulated as follows:

1. What problems do teacher students identify as central to contemporary learning and education?

2. What can their proposed solutions (i.e., co-designed speculative technologies) tell us about how they construe central concepts such as students, learning, and teaching?

3. How can speculative design help us to make problem-solution configurations in education visible, and as such put them under critical scrutiny?

Illuminating and critiquing the connection between discourses and technological developments in education can help to avoid a process of path dependence that can lead to lock-in effects along the way. As such, AIEd mock-ups, role play, and speculative methods open up for ways to explore different educational futures but also truths taken for granted.

Everyday teacher problems

While this study highlights practical issues faced by teachers, it is crucial to recognise that these problems stem from broader economic and political governance of education. Education is often promoted as a catch-all solution to various social problems – such as integration, unemployment, violence, poverty, and poor health – without acknowledging its entanglement with larger social, political, and economic structures. In parallel, transnational organisations have increasingly advocated for detailed governance of teachers’ work conditions and pedagogical practices, often linking them to student learning outcomes (Robertson 2016). These systemic changes have diminished teacher autonomy since the 1980s through reforms that allocate authority to the state, municipalities, principals, and the school market, positioning students as customers (Lundström 2015). Ball (1993) similarly found that various control mechanisms – curriculum, market, and management – redefine the purpose and meaning of teaching, thereby encapsulating and specifying teaching practices.

The ongoing marketisation of education further exacerbates marginalisation and inequity (Bartlett et al. 2002). Consequently, teachers face increased powerlessness and precariousness (Means 2019). Examples include large class sizes (Blatchford and Russell 2019) and budget constraints limiting access to resources (Jackson, Wigger, and Xiong 2020). Teachers must also navigate new educational concepts and paradigms, such as learning styles and paces (Cassidy 2004; Pashler et al. 2008) and nudge thinking (Decuypere and Hartong 2023). Additional challenges include lack of teacher autonomy, accommodating large numbers of students, new administrative tasks, time constraints, parental involvement, insufficient technological systems, and increased workloads due to technology, such as human involvement in automated decision-making processes (Colonna 2024; Sperling et al. 2022). These factors contribute to a complex and demanding everyday situation for teachers.

Speculative methods

The benefits and drawbacks of speculative methods in education research have, to a certain extent, been addressed previously. Mann et al. (2022) presents educational design fictions in general as ‘a generative method for informing and promoting discussion about the future direction of education’ (323). Houlden and Veletsianos (2023) stress the importance of imaginaries of education that are built on optimism and hopeful futures. While my interpretation of Berlant’s notion of cruel optimism (2011) differs somewhat from that of Houlden and Veletsianos (who seem to stress the passiveness of it), I agree that proactively imagining desirable futures is important but also reacting to (and dismantling) the more obstructive realities and their extrapolations. Specifically as, cruel optimist perspectives in education are often fuelled by edtech solutions (e.g., Gulson and Witzenberger 2022; Macgilchrist 2018) and not necessarily pacifying. Ross (2023) and Nooney and Brain (2019) stress working with the interplay of past, present and futures to situate work not merely in reality but also in responsibility. Noteboom and Ross (2024) show how speculative approaches can use (future) uncertainty as a creative starting point, but also that objects-to-think with can help with getting started. In this study, on the contrary, I asked the students to start from a very concrete school-related problem (cf. Levitas 1990), and this difference in approach can probably explain the evil machines that were designed here. As such, using speculative methods can help participants to give an account that is still seen as valid, viable and valuable for their practices, while also not necessarily being optimistic. It is urgent because critical perspectives on, for example, data reporting can shape a deeper understanding of data in education and help teachers to question persuasive datafication and managerial pressure (e.g., Hillman, Rensfeldt, and Ivarsson 2020; Knox 2017). Broadly, visions of the future can be conceptualised as a relation between ‘big futures’ characterised by more controversial, wide-spread and far-reaching scenarios, and ‘little futures’ characterised by more mundane, local and delimited changes (Michael 2017). Similarly, speculative designs can be big or small, with varying impacts. However, this relationship is unstable, as small designs can lead to significant changes, and the consequences of designs make this relationship contingent and mutable. This stresses the impact of privilege and marginalisation for which futures become possible (Benjamin 2024; Harrington and Dillahunt 2021).

Speculative methods and speculative design

On a more design-related level, Pink (2022) argues that a way to engage more actively with futures is through design anthropology, where specific, more or less speculative, designs are studied in new contexts. Light and Akama (2014) stress how ‘[t]he social as it manifests in structure and agency is always political’ (9) and how the feminist concept of ‘care’ can be a way for design to make more long-term ethical and environmental commitments. Mohamed, Png, and Isaac (2020) apply a decolonial perspective to raise questions about how AI systems can incorporate, perpetuate and legitimise structural injustices, including algorithmic oppression, algorithmic exploitation, and algorithmic dispossession. Speculative methods in design research have garnered increasing attention in recent years (Lindley and Coulton 2015). Under the umbrella term ‘design fiction’, several definitions and methods for generating and evaluating design solutions have emerged (Baumer, Blythe, and Tanenbaum 2020). In this paper, I will, for reasons that will become clear, take special interest in what Blythe et al. (2016) call ‘seriously silly design fiction’. The goal of such a prompt is to move away from what has been termed techno-solutionism – a term they attribute to urban designer Michael Dobbins (2009, 182):

The disconnect between problem and solution, always likely to be an issue, became exaggerated in the culture and practice of modernism in city design and planning, where problems were ‘dumbed down’ to meet the solutions offered.

In this quote, Dobbins emphasis problem representations and the important part they play in co-creating a ‘suitable’ solution. What is proposed is simply that the matching of problems and solutions (something that is often emphasised in design) is not a ‘pure’ process unaffected by power asymmetries. To avoid an overly solution-centred design, Blythe et al. (2016) suggest that one turn to concepts that deliberately take into account the unpredictable and paradoxical nature of technology as a way to open up the problem space in question. One such concept, which they point to, is Chindōgu.

Chindōgu

Chindōgu is a Japanese art and design concept, which can be translated into ‘weird tool’. Developed in the early 1990s by drop-out engineering student Kenji Kawakami while working on a mail order catalogue for Japanese housewives, Chindōgu can be described as the art of creating ‘unuseless inventions’ (Kawakami 1995). The term unuseless refers to a paradoxical state where an invention is presented as an ideal solution to a particular problem, but which may in practice also cause more, or other types of, problems. Kawakami and colleagues later founded the International Chindōgu Society, which, holding anti-capitalist and anti-consumerist leanings, in turn formulated ten tenets for Chindōgu:

1. Not Really – ‘A Chindōgu cannot be for real use’

2. Exist-essential – ‘A Chindōgu must exist’

3. Anarchic – ‘There must be a spirit of anarchy’

4. Universally Unuseless – ‘Chindōgu are tools for Everyday Life’

5. Not for Sale – ‘Chindōgu are Not for Sale’

6. Stop Trying to be Funny – ‘Humour must not be the Sole Reason for creating Chindōgu’

7. Propaganda … Not – ‘Chindōgu is not propaganda’

8. Keep it Clean – ‘Chindōgu are never taboo’

9. Don’t get Greedy – ‘Chindōgu cannot be patented’

10. Chindōgu for All – ‘Chindōgu are without prejudice’

A similar effort in the spirit of Chindōgu is Philip Garner's ‘Better Living Catalog’ (1982), which features 62 whimsical ‘life-enhancing products’. Like Kawakami, Garner's inventions blur the line between art and design, and problem and solution. In an ironic mission statement, Garner rejects the idea that technology and automation cause societal problems, instead celebrating the ingenuity and optimism of gadgets aimed at improving life. The catalog’s products are paradoxical and impractical, such as the Digital Diet Loafer, which displays your weight on shoe displays and gives audible warnings for weight changes, adding an ‘embarrassment factor incentive’. While ostensibly solving problems, these inventions invite exploration of overlapping issues like social norms, power relations, and behavioural governance. A key takeaway from silly design fiction and Chindōgu is their resistance to the commercialisation of human-technology relations.

Theoretical departures

As outlined in this paper’s related research section, speculative methods like design fiction and future-making have gained attention for envisioning future scenarios and technology implications. I applied these methods empirically to engage education students in what Daly et al. (2018) term problem exploration, which seeks deeper and more varied meanings of problems. The idea of problematizations, as developed by Foucault (1998) and Bacchi (2012, 2015), provides a vital overlap where theoretical concepts and practical solutions meet. Following Bacchi (2015), I view problematizations through a Foucauldian lens, highlighting that problems and solutions develop political meanings from normalised practices (Bacchi and Goodwin 2016). This stresses how normalised ways of thinking about technology become evident when conceptual problematizations transform into speculative solutions. (i.e., ‘a design’).

Similarly, this aligns with Sicart and Shklovski’s (2020) ‘pataphysical’ approach, which critiques and disrupts techno-solutionism without aiming to implement real-world solutions. Instead, the main reason for the workshops presented in this paper was to explore inequitable impacts of edtech with students, but also to make visible the designed, embedded and systemic politics of artefacts (e.g., Virilio 2007). In certain ways, this is also because an increasing amount of the work that is being done by systems, platforms and technologies is black-boxed or obscured as private intellectual property. This does not, of course, mean that they are any less prone to designer fallacies (Ihde 2008), unintended consequences (Tenner 1997) or system accidents (Perrow 1999). Indeed, as more recent research has highlighted, such contingencies are, due to increased interactive and systemic complexity, perhaps even more ubiquitous today (Agar 2015; Muldoon 2022; Munn 2022; Noble 2018; Reich 2020). As such, the purpose of the empirical work described in this paper is to identify, explore, and counteract an ongoing normalisation and acceptance of technology-centric problematisations (problem-solution configurations) and materialisations. Such a mode of thinking resonates well with ‘anti-solutionist’ and unuseless methods.

Method and material

The empirical material for this paper was elicited from speculative design workshops, in teacher programmes (n = 36), on three different occasions and lasted for approximately 3 h each time. The participants were diversified in terms of gender, ethnicity, and age; however, the respondents are too few for the study to make a relevant statement about any differences between them. The students were given some instructions beforehand and divided into teams of 3–4 people. The collected material consists of 14 design sketches; notes taken during the workshops; plus, debriefing notes made immediately after each workshop. Out of 14 designs, 7 were selected for presentation in this paper since they encompass the full scope of the material without non-intersecting themes.

I ensured students’ consent was obtained ethically, with clear understanding of their participation and the option to opt out at any time. I aimed to create a trusting and respectful environment where participants felt at ease sharing their perspectives and experiences without any pressure. Still, I acknowledge the power dynamics in the researcher-participant relationship. As both researcher and teacher, I hold authority and responsibility. Thus, ensuring the workshop was voluntary, ungraded, and had no negative impact on students who chose not to participate was essential (no student declined or opted out).

The underlying educational purpose and the proposition to the students was to collaboratively investigate how AI in education create, reproduce, transform, augment, transfer, diminish or change power differentials. As speculative designs effectively manifest policies and ideologies in technology, this opens for critique of the politics of AI and autonomous systems in education, as well as how they are connected to oppressive structures. By assigning a speculative design task, and thereby also creating materialisations of educational problem-solutions complex, the politics of its ontological underpinnings becomes visible and possible to question (c.f. Mol 2002). This is essentially to show that design processes involve social choices.

The workshop begun with the participating students being instructed beforehand to think about what they consider is the currently most pressing problem in schools. The teacher students in this study were in the final phase of their education and all have, as the education requires, practised in schools. They were then given time to discuss their problems in small groups, and jointly agree to continue with one of the problems as their focus. The groups were then given the task to design an AI-augmented technology that could solve their chosen problem. The students were not given any restrictions in terms of imaginable materials or technologies at their disposal and provided with large papers and a variety of colourful crayons and pens. Drawing and sketching brought the inventions to life, making the functions of the technology tangible. Students were further assigned to name their speculative design solutions. Naming played a crucial role in assigning specific terms to the initial problem-solution setup. In the last part of the workshop, the drawings of the fictitious AI technologies were put up for display, and I told the students that we were now at an edtech fair where each group could sell their technology to their peers (who in turn had the role of acting as potential buyers). The workshop aimed at conceptual playfulness over precision, creating a non-judgmental space where participants were encouraged to explore ideas, make mistakes, fail, ask (silly) questions, and have fun.

Analysis

The analysis of the collected material took two forms: design critical readings (Bardzell and Bardzell 2015) for the design sketches, and thematic analysis (Terry, Hayfield, and Braun 2017) for the workshop notes.

Design criticism examines and evaluates the values, assumptions, and ideologies in designed artefacts or systems, emphasising their societal, cultural, and political influences. Bardzell and Bardzell (2015) advocate for a broader approach that delves into the implications and meanings of design choices, viewing design as a cultural practice that reflects and shapes societal norms and values. Critical analysis focuses on socio-cultural, historical, and political contexts, underlying assumptions, implications for users, ethical considerations, and power dynamics manifested in design decisions.

After reviewing the data, I assigned meaningful codes to text segments through repeated close-readings, such as ‘performance of irony’ or ‘manifestation of power’. As the analysis progressed, I grouped codes into overarching themes and subthemes, which emerged from the data and were aligned with the study's research questions. These included student-centered learning, teacher roles, technology-as-science, classroom space, techno-solutionism, and power relations. Descriptions of these themes formed an analytical narrative, reflecting the workshop process.

Results: real teaching problems and speculative design solutions

In this section, I will present a selection of the problems and design solutions that emerged from the workshops. The selection of cases was based on problem-solving configurations that effectively showcase the outcomes of the workshop.

Design solution 1. PURR (Personal utility responsive rehabilitation cat)

Identified educational problem: Physical absence (i.e., students who do not show up for class)

Proposed design solution: A robot in the shape of a cat. The robot, carrying the acronym PURR, ‘lives’ in the home of the pupil but is provided by the school. PURR is in constant satellite communication with the school. It regularly ‘scans’ the pupil. After scanning, PURR can, via its tongue, administer a patented solution, which may contain vitamins, hormones, caffeine or amphetamine depending on scan results. It also has claws, which can be used to ‘motivate’ the pupil to (wake up) and get going to school. In certain situations, the cat can also attend school vicariously (e.g., if the pupil is ill, or if the claw-function did not work). Such vicarious attendance must be ‘verified’ – pupils cannot just send PURR simply because they are tired, for example. In these cases, the pupil is also equipped with a USB-port, so that PURR can transfer the ‘learnings of the day’ back to the pupil.

Design solution 2. The planning portal

Identified educational problem: Uninterested pupils who are ‘hard to reach’.

Proposed design solution: Effectively, a door frame that can detect the pupil’s mood and mental status and transfer that information to a ‘didactical conversion printer’. Stored in that device are relevant learning materials for different levels, which has been prepared by the teacher in accordance with the theme for the class. Once the printer has been provided with the scanning results, it analyses these, makes a match between the results and educational content, and prints out a personalised lesson plan for the teacher to use with each student.

Design solution 3. The universal complementary calibrating cyborg

Identified educational problem: Everyone has a different subjective understanding of reality depending on their previous life experience and personal history and as such it is difficult for teachers to fully personalise learning.

Proposed design solution: A robot that will help to adjust and negotiate between the teacher’s and the pupil’s perspectives, so that individualised teaching is made possible. As such, the robot will aid the teacher in reacting to low pupil attention, or to differences in perspective that become obstacles to learning. To achieve this, the robot has both historical and up-to-date information about every pupil and their ongoing ‘life stories’ (by being in constant wireless communication with them). The robot then conveys its analysis to the teacher, who can incessantly adapt and modify their own teaching in order to meet the individual’s needs. The robot would also be able to transform the meanings and verbal intentions of the teacher into a form that is more ‘in tune with’ the world view of the pupil, and thereby more easily assimilated as knowledge.

Design solution 4. Tele-vision

Identified educational problem: Pupils have problems assessing the consequences of their actions. They especially suffer from difficulties in overviewing the long-term repercussions that their (seemingly) mundane choices may have for their future lives.

Proposed design solution: A screen using a form of ‘narrative learning model’ displays how the future will be enacted depending on the choices that the pupil now makes. For example, if the pupil does not do their math homework, they will be shown the impact that decision has on their future selves. Using this technology, the everyday choices will appear more significant and concrete to the pupil, while also illustrating a causal connection between the way we choose to live our lives now and different contingent outlooks on tomorrow (and beyond). Through storytelling, this device will convey both prospective consequences (in a kind of ‘butterfly effect’ way) as well as show pupils how easy it is for them to choose a (good) future for themselves.

Design solution 5. mAIghty

Identified educational problem: Pupils use their phones too much (and for the wrong reasons/purposes).

Proposed design solution: Each pupil is provided with an AI-powered robot seal, which has the capacity to ‘filter’ the content on pupils’ phones and only allows such information that the pupil needs. The seal is also warm and cuddly, creating a sense of security and safety. It speaks all languages and makes sure that everyone’s needs are always met. It conveys religious messages from all religions (although not at the same time, but according to the individual’s preferences), meaning that it can act as a priest, imam, monk, and so on. mAIghty also has a built-in projector that can show teachers what the pupil is experiencing. It knows the answers to all possible questions and can assist in any type of problem-solving.

Design solution 6. S-opt (the schedule optimiser)

Identified educational problem: Lack of motivation. Pupils (and teachers) are forced to do things in ways and at times that are not optimally motivated, which in turn creates negative associations that further lower future motivation for similar tasks.

Proposed design solution: An administrative service that schedules your day in an optimal way. It creates a flow chart based on personal motivation and can thereby condition the individual to arrange their lives in the best way possible. The system creates an increasing number of positive associations to various assignments, making the life of the pupil easier. On a more detailed level, the system includes a complete rearrangement and reordering of mundane life: wake-up services, self-driving buses and drones, homeschooling through the so-called ‘Hallucination’-app, child services, and time management according to a ‘just-in-time’ principle. The system is also synchronised with public authorities, which stops institutions from scheduling (out-of-school) events at inconvenient or non-optimal times.

Design solution 7. Organ-izing gun

Identified educational problem: A lack of active participation in the classroom.

Proposed design solution: A gun that ‘reorganises and connects’ the pupils in the classroom into ‘one coherent body’, that is, one (or a few) pupils become ‘the heart’, another the lungs, another the liver, another the eyes, and so on. All pupils will have to be active for the ‘full body’ to remain in balance and feel good. If the pupil who ‘is’ the eyes is not active, all students will have difficulties seeing clearly. In the same way, the whole group will have trouble breathing if the ‘class lungs’ do not participate actively during the class. In this way it becomes highly tangible, corporeal, and even painful, if all pupils do not collaborate to keep the level of active participation high. It may even lead to the death of the entire class. Through this device, pupils will learn to collaborate and participate equally in a just manner.

Discussion

In this section, I begin by analyse the problem-solving configurations in the students’ fictional designs to show implicit assumptions about learning and education. I then conclude by demonstrating how speculative design can expose and critique the ingrained hierarchies within these assumptions.

Pupils as problems

At the beginning of the workshop, the participants started out with real, serious and self-experienced problems in education (the same way Chindōgu starts out with real problems). These problems were close to everyday practices and returned to issues such as absence, lack of study motivation or difficulties with personalised learning. At this point, however, these problems were rarely questioned in themselves. That is, they were not discussed as potentially oppressive in their ambition to individualise learning or increase motivation as such. Rather, they were presented as part of a well-meaning, and inherently good, discourse. Once implemented and operationalised in technology, the oppressive powers became too clear.

One reason for this is although the problems concern the teacher’s difficulties in everyday practices, the design solutions target the students aiming towards getting a better (but in certain ways, also limited) understanding of pupils’ dreams, hopes, and ambitions. The scientific measurability of previously hidden parameters becomes the grounds on which personalised learning is based. Essentially, it is the sender-receiver relation that is being optimised (bio-technically). All the solutions also express a great confidence in measurability and/or surveillance. The fictive machines are thus seen as capable of revealing a ‘truth’ existing underneath or beyond what is humanly perceptible. The logic is that since teachers cannot see how things ‘really’ are, teaching and learning will potentially fail. The goal of reaching that imagined potential of education is superior to (any consequences of) the means of getting there. For example, during one of the sales pitches, one group received a question about surveillance: ‘But what if the pupils don’t want to be watched all the time?’ To this the group replied: ‘Not to worry, this tech will be completely hidden and undetectable’. Another student pointed out that they already have daily so-called ‘well-being check-ins’, when pupils are given the opportunity to say something about their mood and current state of mind. In reply to this, the technology was still made relevant since ‘it could inform the teacher if the pupils were really telling the truth’.

The proposed design solutions exemplify prevailing discourses surrounding student-centred and personalised learning standards. The issues are frequently characterised as difficulties in engaging pupils at their level, primarily because the pupils’ lifeworld remains invisible to the teacher. In other words, the teacher lacks comprehensive data about the pupils, and to achieve a fully optimized and individualised learning context, they must first obtain ‘real’ and ‘true’ information on which to base their instructional decisions. Consequently, the challenge posed by pupils’ varying levels of ambition and prior knowledge can be reframed as the teacher's lack of understanding of the pupils’ specific needs and the underlying issues they face. As seen, this can be solved through a ‘scanning’ of the corporeal status of the learner. A related problem can be found in that pupils are agents in different and overlapping contexts, many of which the teacher has no control over. Therefore, the resulting design solution expands the teacher’s authority to encompass the entire lifeworld of the pupil. Similarly, a lack of motivation or comprehension regarding the potential benefits of education is viewed as an obstacle to learning, necessitating the adjustment or ‘correction’ of the student’s emotions and attitudes. The solution to this problematization involves biotechnically nudging or altering bodies and minds to promote conformity or using a medium to translate and standardise teaching and feedback (c.f. Decuypere and Hartong 2023).

In a way, this illustrates a desire for, or at least acceptance of, a fully automated society, where individual duties and responsibilities diminish (through automation). Throughout history, the idea of a fully automated society has been repeatedly invoked as a utopian dream (e.g., Hong 2021) and is seemingly still so. Even though ‘there is now an app for everything’, all the promises of smart technologies that would revolutionise education have not yet been fulfilled. Thus, there is still a large design space where speculative solutions can satisfy (ironic) dreams of education automation, and at the same time also question both inherent power dynamics in technological solutions, as well as the need for automation to begin with.

What are schools for?

The problem-solution configurations manifested in these speculative designs make ideas of both teaching and teachers visible. Student-centred learning is emphasised, but so is an ambition to retain the teacher at the centre of learning. It would be possible to ‘design away’ the teacher, but the teacher is instead given an extended role as an interpreter of pupil data and (sometimes) also as a transformer of data insights into teaching practice. In that sense, the proposed designs seem very sensitive to the ‘original’ ironies of automation (Bainbridge 1982). This means that the proposed designs both provide the teacher with data, as well as help the teacher to shape the pupil as someone who is ‘ready-for-learning’. This puts considerable responsibility on the teacher but also seems to promote certain theories of learning over others. Thus, these fictitious AIEd technologies ‘solve’ the difficulties of adapting to each pupil’s needs in three ways: (1) by providing the teacher with previously hidden information about pupils (e.g., their moods, their life stories, their learning styles, their contexts); (2) by mediating the teaching/learning itself; and (3) forming the pupils will towards a desirable ideal. As such, the speculative designs work through the adaptation of both the students to learning, and the learning to students. What is also clear is that the technology becomes a neutralised (‘truly scientific’) medium without any agency or initial politics of its own. Instead, it supports an increase of individualisation in the classroom, which upholds the separation between school, individual, and system. Surveillance is simultaneously a prerequisite and a solution. Paradoxically, the individual learning styles, backgrounds, and other characteristics of students are often addressed in ways that ultimately reintegrate them into the conformity of the classroom. Thereby, student-centred learning becomes an ironic practice, as it is the process by which a standardisation of learning can be achieved. The speculative designs reproduce the pupil as passive (although passiveness is the original problem), and pupils are thus seen as recipients at the end of scientific governance. All the speculative solutions retain the teacher as someone who is central to learning, and someone who should not be replaced by technology (rather their role is re-centred through technology).

The speculative designs also cling to the classroom as the primary place for learning. Some of the designs include possibilities for mediated learning, but still, learning is seen as mainly happening in the classroom.

Confronting evil AI

It was mainly during the final phase that the politics of the designs became apparent, in what Light (2021) describes as a ‘result from designing both a provocation and a process for encountering it’ (1). During the design task itself, the students were so busy solving their selected problems that they rarely had time to fully discuss its problematic aspects or consequences. Once they began to convince their peers, the more problematic issues begun to unfold. The ‘selling group’ often performed an ironic imitation of a techno-solutionist discourse, which effectively glossed over and whitewashed all doubts and objections, such as exaggerating the ‘commercially oriented touting’ of the invented solution. For example, one group (having given their solution the shape of a gun) were asked if it would not cause unease for teachers to ‘shoot their pupils’, to which they responded: ‘no, no, this is a friendly gun!’, which in turn generated much amusement among peers. Exchanges such as this were very common during this phase, and the discussions, at times, reminded one of a heated, but friendly and humours, debate. This clearly resonates with the Chindōgu approach, where the ‘unuselessness’ of designs become a way to highlight how solutions may generate new problems. It also echoes several of the specific tenets, such as tenet I (‘it cannot be for real use’), III (‘a spirit of anarchy’), and VI (‘humour is not its sole reason’). Notably, it also confronts some of the Chindōgu tenets, primarily V (‘it is not for sale’), where the process of selling it becomes an important reflective moment in these workshops. The participants’ designs are also not without prejudice (Chindōgu tenet X) – this was something that ‘just happened’ in the process – but this ‘failure’ became an important prerequisite for reflections. Thus, when the participants were forced to solve their problems via technology, the governing aspects of both the problems and solutions became visible. As one student said, laughing: ‘But still, this is what we would actually like to do!’. This comment makes visible the paradox of teachers’ everyday life living up to sometimes impossible ideals, as well as the paradoxes inherent in Chindōgu designs. Also, the obvious Chindōgu silliness of the solutions became apparent to everyone, but acting like they are as normal as anything became an ironic performance of edtech solutions, and perhaps technological solutionism at large, which is also clearly a critical stance, questioning the always-supposed usefulness of external quick fixes. As such, the final phase of the workshop became a combination of presentation and debriefing, where the evil designs triggered a mix of performance and debate. The final phase is therefore critically important as it collectively (e.g., Benjamin 2024) opens up to resistance and disruptions both against non-useful edtech solutions in themselves, but also against an individualised prompt to provide solutions to structural problems on your own. A common problem is otherwise that participants in speculative workshops often neutralise, naturalise, and legitimise certain technologies and values that ‘discursively close’ (Markham 2021) rather than open discussions about alternative futures.

The excited and jaunty atmosphere in the classroom, when the students start joking and mocking each other about their inventions, indicates that the teachers-in-training realised perfectly well that they had created very authoritarian and oppressive edtech. An example of that is when a group of students tried to sell their technology to the others, I asked if anyone could see any power asymmetries or oppression in the design, whereupon the whole room burst into laughter in response to this too obvious question. That, combined with the fact that all speculative designs without exception were problematic and the amusement that occurred when the students saw the craziness of their peers, indicates to me that these ‘silly’ and ‘failing’ designs created a common understanding of the problems that these speculative designs highlighted – problems that also resonate clearly to real technologies and experiences.

Conclusion

This paper contributes to moving forward current discourses about speculative methods by explaining how they can become critical in practice. It further reveals how ideas about good learning or desirable education are dumbed down in technical solutions, different types of oppression, power asymmetries and surveillance arise. As such, it illuminates the underlying mechanisms driving dissemination of technology and explains why it can be harmful. However, these exercises of power are not created primarily by the technologies but by the institutional, discursive and economic framework surrounding the school where teachers become bearers and realisers of the school’s goals. What appear to be the good goals of the school can quickly turn into mechanisms of oppression and coercion when translated into technical solutions.

Speculative methods and imagined solutions need to go through an ‘ironic’ phase in order to realise their critical potential. Without a reflection upon the ‘silliness’ or ‘evilness’ of the solutions, there is a pertinent risk that reflections are halted in restricting imaginaries, including techno-solutionism and uncritical technology acceptance, or, in a worst-case scenario, just identifying new markets for edtech companies to exploit (e.g., Good 2021).

The oppressive hierarchies that arise in the students’ fictitious AIEd reveal how the seemingly noble goals of personalised learning and the teacher's role as a learning facilitator harbour oppressive aspects inherent in the education system. These technologies make visible how biopower, as described by Foucault (2008), operates through technology schools. According to Foucault, the institution of the school, laws, economic conditions, and curricula can be understood as techniques of biopower, while educational technology is the disciplinary technique that produces data as a power/knowledge regime. This highlights the importance of recognising educational technology’s intimate connection with the school’s overall role in shaping citizens. Ironically, even the best educational discourses can create deeply oppressive orders because discourses constitute practices (Bacchi and Bonham 2014). Unanimous understandings of what constitutes good schooling, or a desirable educational future make it difficult to question the power asymmetries involved. Concepts like personalised learning, student engagement, inclusion, or individualisation are often considered inherently positive, obscuring their potential for oppression. This case study demonstrates how educational goals that are taken for granted as desirable can contain oppressive elements that surface when concretised in technical solutions. Speculative methods can thus help uncover the oppression embedded in unanimously accepted notions of desirable educational futures. As such speculative methods open for a problematisation of education problems and imaginaries. That is, the proposed technologies address certain types of problems, which have become ‘problems’ because of certain views of learning and teaching are normalised and neutralised. The problems initially identified are ‘imposed problems’ relating to the division of labour, curricula directives, neoliberal framings, or management issues that are pushed down to teachers to handle in their everyday practices. This is also indicative of how teachers find themselves in a position where the dominant logic of individualisation and personalisation (Selwyn 2021) are not only applied to students, but also to themselves. For instance, if students are missing from class or not engaged, teachers are expected to solve these issues. To be prompted to think ‘futuristically’ often results in extrapolations. As such, speculative methods reveal current discourses by operationalising them, and thus the power asymmetries and exploitations are made explicit, from which actual critical speculation can emerge.

There is an often-repeated statement that neoliberal society has lost its capacity to imagine other futures beyond capitalist systems (cf. Bloch and Adorno 1988). Paradoxically, the proposed solution to the inability to imagine non-capitalist futures is to engage in reimagination and utopian thinking. Responsibility for the future is thus placed on the individual’s imagination and will. The solution to contemporary problems is construed as the ability for positive thinking. In this way, the future is shaped as a problem, and the solution is created as a manifesting programme with the aim of making the individual willing to engage in the creation of the positive future rather than be critical of the present. Negative expressions such as protests, resistance, or passivity represent subjective positions that are challenging to sustain when attention is solely on the positive. This focus on positivity also overlooks the potential need to destabilise, dismantle, and oppose current and future forms of oppression and injustice. Moreover, apart from this requirement to create a positive future, teachers may also be unable to function as critical watchdogs against an eager tech-industry if achieving quantifiable educational goals becomes more important than the just way there. They risk forming a bio-socio apparatus that, while becoming more efficient through automation, fundamentally derives its power from prevailing discourses.

Speculative methods can thus be productive in making visible power asymmetries in taken for granted discourses about desirable education and as such function as ‘a chemical catalyst so as to bring to light power relations, locate their position, find out their point of application and the methods used’ (Foucault 1982, 780). This study has shown that common goals for education could become evil when operationalised in autonomous technologies, because algorithmic control works according to the ends justifies the means logic. Fortunately, the technologies here are fictional, as Sian Bayne (2023) stresses, ‘[w]e hold it in our power to become differently technologised’.

Acknowledgements

My warmest thanks to all participating students who, through their strong commitment to social justice, humour, and courageous self-reflection offer more imaginative, more cooperative, ways of being. My sincerest thanks to the anonymous reviewers for their helpful generosity.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was partially supported by the Wallenberg AI, Autonomous Systems and Software Program – Humanities and Society (WASP-HS) funded by the Marianne and Marcus Wallenberg Foundation and the Marcus and Amalia Wallenberg Foundation.