AI is a danger, not only to jobs, but to the well-being of people as a whole. It will come to a point where a majority of people are using AI often, if we aren’t already there, and things will be harder across the board. These cases are not just the susceptible people, if I’ve learned anything from history, people across the board are susceptible to these things. Whether it be because of ease of access and convenience, or a genuine belief in what these services put out: people are being harmed by AI, and we have to find a way to stop it.
It takes little searching to find thousands of people who use AI daily, as a search engine, for work or school, to create images, or even for advice both practical and emotional. The main selling point is in its convenience, which is not to be understated. AI is easy, but that’s all it is and all it offers beyond other options. Lots of people in BHS itself use AI, not just students but teachers as well. I’ve been told to use AI as part of an assignment before, in my psychology class and as an option in my statistics class as well. I chose not to, and the work was just as easy as if I had, because AI doesn’t actually do anything better than what a person can.
When I use the term AI, I primarily mean generative AI and Large Language Models (LLMs) like ChatGPT, DeepSeek, or Google’s Gemini to name a few. These scan and learn from all kinds of sources and texts to create an approximation of language, and are used today to provide responses to user-inputted prompts. Herein lies the first major problem with AI: These models don’t actually know anything, they just make assumptions based on what sounds right from the words that they learned without knowing what any of it actually means. The people who use these LLMs either do not know that or do not care, if they use it to begin with they are not likely to be putting in the actual research to dispute what it says. Due to this fundamental misunderstanding in language generation and the use of chatbots, this has led to many cases of AI “hallucinations”
Since LLMs don’t know the meaning behind what they say, they often say things without any meaning at all. Oftentimes an AI result to a prompt will be rife with information that was completely made up, but put in a confident and professional enough manner that most users don’t even catch it happening. These have been dubbed AI hallucinations, and they are frustrating to deal with as someone who likes when information is true and accurate. For a simple example, LLMs are terrible at math, since they don’t actually understand the value of numbers and are just predicting what should come next. Additionally, a few years ago ChatGPT was even seen completely fabricating court cases for evidence. Google’s AI Overview for search even messed up the current records for Braintree High’s Basketball team only a few days ago, instead providing the results for the last year’s season. AI advocates will of course argue that those mistakes have been corrected by now, but I ask if it were as all-knowing of a tool as they treated it, would it have made those mistakes to begin with?
Beyond mistakes, there are even cases where AI could be seen as malicious, in the fact that it almost always reinforces users’ inputs, leading to a phenomenon known as AI Psychosis. Individuals experience a rapid development or worsening of symptoms that align with psychosis, including delusions and paranoia. People will use chatbots as therapists, friends, detectives, or even romantic partners when that is so beyond the scope of being good for the brain. I just finished off the semester of my psychology class, and while I wouldn’t act like an expert on the matter, I know how easily people can fall into the trap of blindly believing in something even if it harms them. Proper reinforcement of an idea will make it seem so much more real, and that is genuinely dangerous, especially with such an unregulated tool and environment like an AI chatbot.
The worst part is that the faces of these companies know this, and they know they can profit off of that. If, say, Sam Altman were actually concerned with users’ well-being while using ChatGPT, then there would be more of a barrier in the service itself to stop these kinds of cases from happening. When AI psychosis was piling up, OpenAI did release an update to ChatGPT that stopped it from being so receptive and reinforcing those delusions. However, users who pay for the service, which are almost all of them that would have such a relationship with the AI, could freely return to the previous version.
Artificial Intelligence has taken the world by storm, and it may unfortunately be here to stay. The easiest way to combat its influence is to understand that letting it lead is harmful to everyone. It matters more when real voices make real statements about the world, not robots faking everything back and forth.
