I’ve been recently studying criminal profiling to upskill my court-expert practice. I get material to study, to do homework, material to practise, obviously – like any other student at any other school.
I also research stuff on my own – and you can see the results on this website, in the whole ‘Writing’ section.
What I’ve noticed lately is that there are more and more topics that seem to be forbidden to dig deeper into when using AI – even just for learning or researching.
The first one is kind of obvious. Criminal profiling isn’t a topic AI is willing to help me learn. Tools like NotebookLM (now Gemini Notebook) don’t even want to accept this kind of subject, even when it’s explicitly for educational purposes. Now I’m not even allowed to research the topic among academic papers through these platforms, so I’m forced back to doing it manually.
I was also flagged by ElevenLabs. Thank goodness that one was resolved quickly (thanks, ElevenLabs). Anyway, flagging me for violating content rules when I’m just trying to process my own study materials was a surprising experience. This one I kind of understand: criminal profiling is a set of difficult topics, not a typical small talk matter – fine. What I don’t understand, though, is how it’s even possible, when we praise AI so much, that it still can’t differentiate between researching a subject and actually executing it. How come that neutral academic investigation gets caught in the same filter meant to block hate speech or stereotyping? How is it even remotely possible that we still have the gap between intent-aware understanding and pattern-based content moderation? Just this time, it’s not a technical limitation…
But what struck me even harder was that topics like victim mentality, its cultural designations, and symptom research are no longer things Gemini wants to help me with. Thank you, Google; I really appreciate your deep concern for my reading list. Apparently, when I want to do sociological research on victim mentality based on cultural descent and the history of specific ethnic groups – comparing groups, not targeting any single one – that’s suddenly off-limits. Having had the outline for my book on victimhood and victim mentality in my mind for quite a few months, I am now officially back to doing my research manually. At a time when everyone thinks AI is about to automate our jobs and take over high-level work, I severely doubt that’s going to happen anytime soon, not when AI starts to be a heavily gated search interface itself.
We are in the 20s now, so maybe it is time for yet another type of prohibition!
Because after all, what if we all start researching difficult topics? What if people start asking uncomfortable questions? Who will then be able to steer the public’s viewpoints? And the most scary – for me – question of all: will the good old search follow the same trend? Is it the time to renew my academic library card now?