What if I began this article with a disclaimer that it had been partially or fully written by artificial intelligence (AI)? Would you trust its content? What preconceptions would you have about the source of information, written word, or author?
I remember the first time I was in a computer science class that permitted the use of ChatGPT, one of the most prominent large language models (LLMs). The syllabus naturally contained limitations and regulations—most notably, the requirement to cite any AI usage—and a reminder of Vassar’s academic integrity policy. Throughout the course, I remained skeptical of the ethics of using a tool I realistically knew nothing about. My first experimentation with the AI ended up being the ultimate assignment for the class, when we were instructed to use ChatGPT and comment philosophically on the tool.
As ChatGPT continued to gain traction throughout 2023, the ethics of its use slowly crept their way into daily discourse. Last semester, I had a conversation about the growth of generative AI (genAI) tools, the umbrella term for AI tools including LLMs. The individual I was chatting with proposed the idea that the skepticism around AI tools is rooted in a lack of knowledge about them and fear of their power, much like the perception of search engines before they became commonplace. However, the previously nerve-wracking technology has now become a tool for daily use, as is currently evidenced by the option to read this article online.
I was born shortly after the internet became popular, so learning to use computers and search engines was a facet of my elementary and middle school education. Once a week, we went to the computer lab and completed typing practice and various tasks outlined for us by our teacher and a lab specialist. I often enjoyed research-based tasks, when we were given a topic and told to find relevant information or sources for short reports. However, even as young as sixth grade, we were warned about the vast arrays of misinformation that existed on the internet. We were given lists of databases that contained reputable work, and taught how to identify reliable sources we found independently. Most importantly, we were shown how to recognize bias and how to acknowledge it in our own work.
When it comes to technology, the generations prior to ours are often seen as unwilling to adapt to or learn how to use new tools. I imagine that, had I been born earlier, I might have shared their skepticism of the internet and its uses. However, I was fortunate enough to grow up in an environment that instead fostered healthy curiosity about technology, and I believe that doing the same for AI tools will help build AI literacy.
This summer, I had the opportunity to work in the GroupLens lab at the University of Minnesota, Twin Cities. In the initial phases of my internship, I reviewed literature about overreliance and mistrust in AI tools. A common theme across the articles I read was that individuals who had previous experience with AI tools were more likely to question AI outputs in a given task. In general, overreliance—or blind trust—originated from a place of inexperience. Learning about the current problems facing the industry and what is being done to remedy them largely helped me focus on what I needed to learn about AI tools to make myself more comfortable using them.
Tools like ChatGPT are here to stay, and it is important to recognize their growing permanence. Instead of stigmatizing their use, we should focus our attention on creating “best use” policies that are unique to the tasks they can complete. In the media, I have seen “AI cheat sheets” listing the best queries for obtaining sensible information based on what you want to know. These are great resources for beginning responsible experimentation with AI, and more people should be aware of them. Like typing and online research, AI should be taught in a classroom setting.
Another common concern is that AI will eventually replace humans. On the contrary, learning about AI can help keep its use more human-centered. By encouraging humans to work in tandem with AI, we can create environments where contributions from both are equally necessary and beneficial.
For the record, this article was not written by AI, but it just as easily could have been. There is a growing body of research on the power of genAI tools, and it is only a matter of time before they progress into consistent use. Instead of viewing genAI as an autonomous monster, consider learning more about these popular tools, and how they can be ethically and responsibly integrated into daily life.