Here we come to perceptions. A well-versed reader of the Think Tanks needs no introduction to this probabilistic inferencing process that is a whole-brain state. But considering the wide scope of readership, let’s proceed with a very brief idea of the perception process.

The experiences of everyday life provide us with a wealth of signals—of a sensorial and interoceptive nature or, to say it another way, from the external or the internal environment—that allow for a computational process of matching incoming data to the internal model of the world (yes, perception starts on the inside!). Such computations—or predictions of what the new signal might mean, based on previously encoded sets of data (our previous knowledge)—are used to make meaning from the new signals. If there is a mismatch, or a prediction error, information about the mismatch also gets encoded in our internal models. The end result of this [never-ending] process is a perception (Barrett, 2017; Friston & Kiebel, 2009).

Now, let’s imagine the perception process of a teacher who has issues of their own with technology. Be it because they belong to a pre-AI generation or have had bad encounters with technology before, generating anxiety or discomfort, or because they have no time or resources to engage in a new learning cycle (about AI literacy, for instance), the end result, i.e., their perception of AI, will be biased. Not in a positive sense.

That said, let’s move on to the next stage, well-being. In post-pandemic times, we are all very aware that mental, emotional, and health factors combine in the delicate balance that guarantees a positive feeling—a working definition of well-being. That balance may be disrupted for many reasons, mostly those that concern our everyday lives, such as work conditions. Let’s list some: (1) When teachers are constrained for time, as when they have little time to plan and make do with curriculum requirements (Chen et al., 2012); or (2) When they lack support (a) for spending time with peers in social interactions (Carew et al., 2020) that provide discussions of what constitutes learning or (b) from leaders who do not understand that professional development is a requirement rather than a luxury (Casal-Otero et al., 2023). The result of all the above causes is the same: Teacher well-being is the first thing to go.

Now, when well-being is affected, teachers may fail to adopt and use AI. This happens mainly because AI adoption and use rest on a decision-making process that demands a lot of brain resources. Although the decision is a conscious result, the process of making it happens mostly at an unconscious level—for metabolic budgeting is not like calorie-counting—plenty of resources are needed. As teachers’ brains, like all brains, are experts in body-budgeting to guarantee well-being, resources will be needed. That means that teachers, when faced with deciding upon AI, will engage in an internal dialogue that consists of a trade-off, which may run in the following way: How easy is it to use this new technology? What are the metabolic costs of accepting it? To what extent will it help me do my job?

Such a decision-making process—again, a whole-brain endeavor—is wearing and tearing. After all, teachers’ brains are processing principles, applications, ethical considerations, and security measures that form the bulk of AI literacy (Casal-Otero et al., 2023; Walsh, 2017). Therefore, perceptions of ease of use and usefulness of AI together with user acceptance (Davis, 1989, 1993) may get up-ended if well-being is compromised. It’s time we examined AI.

AI is a computational system designed to simulate human thinking processes and behaviors. Now, when there’s a human on the scene, bias will be there. And bias gets into algorithms, jeopardizing design thinking. That is misdesign, a plague that needs to be scorched out of AI systems by means of transparency and fairness, utmost aspirations that walk hand-in-hand with equality and regulation. Another issue that corrupts AI use is misuse. An example: When teachers, instead of thinking about the ethical use and pedagogical appropriateness of AI adoption, use an AI tool unthinkingly (much like they did when posting pictures of students online), they may risk multiplying the inequalities that plague most of AI, which is unregulated  (Zinn, 2021).

Teachers—and other educational agents—need to understand that accepting AI is different from preparing to use it reflectively and ethically. To that end, they need to be aware of what objectives drive their use of AI. Further, they need to make sure that such objectives align explicitly—and in unequivocal terms—with the AI tool(s) they are about to use. In short, they need to know how data (“the what”), algorithms (“the how”), and models (“the to what end”) are put together.

If such literacy does not happen, followed by specific training, with adequate monitoring and support from other stakeholders, teachers will feel the pull of adoption without ethical use. This means an unbalanced trade-off that reduces their well-being. In a state of unbalanced body budget, perceptions will become distorted and affect—negatively—attitudes and dispositions towards AI (Ramacciotti, in press). That’s a recipe for a disaster we can prevent (Delello et al., 2025; TeachFlow, 2023). AI is everywhere, but so are we. As humans, we still hold power in how we perceive, relate to, and work with AI. The force is—still—with us.