I believe artificial intelligence is stupid.

Not in an insulting way, but it really is. For example, I tried asking Google’s Gemini what foods are simultaneously banned in America and legal in Europe. The answers I got as of Sept. 8 included foods that are banned in Europe, but not in America.

Even asking the free-to-use model of ChatGPT will yield similar results, including answers such as, “Azodicarbonamide…legal in the U.S and banned in the EU.” The result it gave is the exact opposite of what I asked, and I believe this fundamental misunderstanding by artificial intelligence (AI) actually reveals a major flaw in the design of most artificial intelligence as of late.

While most publicly available AI systems are designed as language models, the information they generate is often skewed towards commonly searched terms, which quickly creates echo-chambers of biased information, regardless of user bias. Among other issues, like certain models hallucinating verifiably false information and reports of some models encouraging self-harm, I still believe that the most significant threat AI poses is in the academic setting.

This past year, CSUN declared itself the first AI-Powered University System in the US. While this may seem non-threatening on the surface, after considering it for more than a few seconds, I believe multiple problems can arise.

First of all, the only stated change is that students and staff will be given access to ChatGPT Edu, a version of ChatGPT. I believe this awfully generous offer from OpenAI has a glaring oversight.

As far as I’m aware, we do not have any mandatory or even optional classes at CSUN on how to properly use AI. For a school that wants to embrace the future, I believe not preparing students to use such a versatile tool seems a bit irresponsible, especially when the mandatory General Education (GE) classes often require students to learn how to use archives like OneSearch for finding reliable sources.

Something that has been drilled into my head over the duration of my academic career is how bad plagiarism is. With the growing popularity of AI, many professors seem to have lost hope, with one of my own even stating, “Students will just use AI either way.”

In my opinion, there doesn’t seem to be a consistent policy on AI, even within the same departments at CSUN. For this reason, I suggest, along with a required segment in low division classes on proper and ethical AI use, that we treat AI like we do Wikipedia.

A publication from the Harvard College Writing Program discusses an advisory that instructors have on using Wikipedia, which I believe sounds familiar to the warnings some of my professors have given me about AI.

“When you’re doing academic research, you should be extremely cautious about using Wikipedia,” said the Harvard College Writing Program. “If you do start with Wikipedia, you should make sure articles you read contain citations.”

I believe applying this methodology to AI is the absolute simplest and most efficient way to outline how students should be using AI in any academic setting.

In my opinion, AI is inherently unreliable, prone to creating echo-chambers and should never be trusted unless the user can find a reliable source to support the AI’s claims. At which point, I believe they shouldn’t even be citing the AI, but the better source instead.

The extensive use of AI is an existential fear of mine; allowing machines to constantly summarize readings and complete academically rigorous work for students will lead to graduates who are less qualified than ever before.