In the rush to integrate Generative AI (GenAI) into education and research, effective and responsible use has largely been treated as a technical challenge and a matter of writing better guidelines, refining prompts, or choosing the right detection software. However, current research, including ours, suggests that the path to responsible use is profoundly impacted by our emotions.
The integration of GenAI tools into research is becoming more widespread, yet it remains largely invisible. In a systematic analysis of 3,384 research articles in applied linguistics, we found a noticeable ‘transparency gap,’ where only 2.8% of authors disclosed their use of GenAI (n = 94). This low disclosure could be because of the emotional hesitancy authors feel and the fact that reporting conventions are still emerging.
In an academic culture that prizes original human thought as the ultimate currency, using GenAI can trigger shame, and a sense that disclosure may make scholars appear lazy or threaten their agency; combined with uncertainty about their competency to use these tools responsibly, this transparency paradox can produce emotional hesitancy around declaring or even using the technology in the first place.
To understand these emotions, our research turns to Pekrun’s Control-Value Theory (CVT). When a researcher appraises AI as a threat to their agency or doubts their own competency, they experience low control, which can lead to deactivating states like shame. This shame then discourages disclosure unless such reporting is explicitly encouraged or required. Conversely, when a researcher views the tool as a manageable assistant with high value that falls within their competence, their emotions shift toward activating positive states like curiosity and pride. In this state, a researcher might feel empowered to be transparent.
In our earlier research, we proposed a five-component framework for responsible GenAI use in research: transparency, methodological rigor, ethics, societal value, and human accountability. However, I argue here that this framework cannot be treated as a mere checklist. Each of these components is deeply influenced by the researcher’s emotions. Emotions act as the hidden layer of the framework; for instance, the decision to be transparent or the confidence to maintain methodological rigor stems from a researcher’s sense of agency. When a researcher feels a strong sense of control and value, these five components function as intended. On the other hand, if a researcher feels that GenAI threatens their identity or outpaces their competence, the framework may be less effective. They may then hesitate to disclose their use of GenAI because of apprehension about how such disclosure will be perceived.
Ultimately, to ensure the integrity of our field, we need to move beyond technical tutorials and pay much closer attention to the emotional experiences of both researchers and teachers as they navigate this technological shift.
