As of 2026, AI tools have become more prevalent than ever before and are being actively incorporated into all everyday applications. From customer service, to simply composing an email for work, AI components have been incorporated into almost every facet of the modern age online. While these tools seem to be great, one of the shortcomings of being in this new “AI Age” is that there are many misunderstandings and inaccuracies caused by these generative tools . To test how viable and accurate these tools can be for essential tasks, I performed an experiment by generating a blog to test the factual accuracy and summarization of ChatGPT.
To begin my test, I simply asked Chat GPT to write a “mock” blog post without citations comparing the operating systems of MacOs and Linux. While not exactly the most specific topic, my goal was to test for accuracy and determine if the average person can rely on AI generated information. In the initial inquiry, I asked for this pseudo blog post to be approximately 250 to 300 words and as accurate as possible without citations. After reading through each line, the results were interesting. From what I could tell, the AI model, specifically ChatGPT was actually quite accurate in it’s summary of comparing MacOs and Linux. While it wasn’t exactly particular, it still was useful at explaining to the user the benefits of each environment. However, a phenomenon I did notice is that there are some phrasings I found to be slightly misleading for someone who is familiar with each of these respective platforms. So while it can give a user a good idea on the “big picture” so to speak I would definitely research more on reliable sources about any given topic as opposed to just asking ChatGPT.
So to summarize, what is my opinion on the experience? To be honest with you, I think LLM’s such as ChatGpt are essentially becoming modernized encyclopedias. They can give the average person a good idea of what a concept or topic is, but not necessarily the most accurate “nitty-gritty” details. I believe that this will change with time, and possibly already has, if I used a subscription based model instead. Nonetheless, I would always be sure to double check sources when reading a statement/excerpt made by AI. One of the easiest ways I do this is by simply asking ChatGPT to provide a source for the statement so that way you can see where the AI is drawing the primary information from. I believe that while there are now still small errors in grammar, writing, and factual information, these models will quickly improve with time and one day most will not be able to pick out any flaws in generative media.
By the way, here is my google drive document where I compared, commented and edited the generated blog post!
Nathan's Blog 