I recently had the chance to watch an episode from X_Files, which shows an AI gone rogue (S1, E7, “Ghost in the machine”). The story shows an intelligent operating system becoming sentient, and taking actions for its survival. I do not remember how I felt when I watched this when I was younger (at least 10-15 years ago). The show itself was a favourite of mine, and somehwat inspiring too. But, this time, as I watched this with my wife, it was only a bit amusing to me. Nevertheless, we agreed that it must have felt very differently to the people back then. Did we find it only amusing because AI itself has become more familiar to us? Or is it that we downplay such scenarios involving AI? or is it because we already know that it is a possibility? Only weeks ago, HuggingFace hack happened (https://metr.org/hugging-face-incident-report-aug-2026.pdf). Frequently, researchers state the risk of AI becoming rogue, and perhaps threatening humanity. Some argue that such threats are not credible and current AI is merely mechanistic and dumb. Especially, some argue that most of these (including other claims about present AI capabilities) are only marketing tactics, inflating the facts and perhaps nothing other than result the irresponsible practices of some AI companies (https://www.youtube.com/watch?v=54Lu6_ZRF0c). As I think, If we look beyond (or maybe not beyond, just nearby) this possible threatening nature if AI, we may see that AI poses a different kind of dilemma, which is our ‘dependency’ to it, which could be a very costly one. I believe all these ideas have already been discussed or argued about, but here are my two cents. Throughout the history, many dilemmas that the humankind faced have been of our own making. Wars and their implications are obvious examples. Certain Non-commucable diseases (except for unfortunate causes such as genetics) are another example. The unprecedented implications of AI could be just another similar example in the future. One way it could happen is due to our dependencies on it. Currently, such dependency has been created, and being reinforced as the time goes by. As intellectuals argue this may or may not create other associated problems, such as a genration of people who do not think well (cognitive dependency), the resource consumption etc. Yet, it is really difficult to get away from the dependency as well. I think the primary reason is that the efficiency boost that AI is capable bringing into economic activities does not give an incentive to turn away from it. Only a bigger incentive such as preventing a risk to human safety/health would do otherwise. Another reason for dependency is that the values individuals attribute to certain AI systems (e.g., chatbots that are seemingly intelligent), as if they have intelligent capabilities. This masks possible blind spots that they obviously possess, and perhaps strengthen with the time, and more interactions. Perhaps, just as social media/short content cater to our boredom, and anxiety in a way (and thus woresening it with time), AI might be doing that in an unseen way. Yet, as humans try to adopt AI in safety critical tasks (such as autonomous tasks that is already using AI) more and more, there may be better chances to realize and solve those problems. On the other hand, people do not learn lessons from their past mistakes (not true to every case, but surprisingly at some disastrous cases. e.g.- wars), especially when it involves attracitve incentives. The disruptiveness (or distructiveness) of some modern age decisions are only coming into light now (unfiltered digital media and short content). Should there be a major fault of AI that pose a visible threat to humans collectively, so that more eager guardrails would be applied, and make humans rethink about the dependency to AI? or would such a case still give power to a limited set of individuals? In a next post, I’d like to think about what long term implications, interventions should there be regarding AI adoption.
Update (19 Oct 2026): I did not that there was a paper presenting a mathematical model of LLMs as a cognitive virus (https://arxiv.org/pdf/2609.03344). I think this is a reference to what was mentioned earlier, where AI behaves in a similar way to Social media ( in terms of cognitive impediments).
This post is licensed under CC BY-NC-SA 4.0. The X-Files still is excluded and remains the property of its copyright holder.