The general caveat with using Generative AI for anything related to academic and research work is to approach it like a powerful tool that has limitation and works best when it has substantial user oversight. Like any tool. it is not a replacement for your own research and work. Not only should good academic work involve multiple sources and double-checking, but there are issues with academic integrity if work is passed off as your own (as the technology continues to be more widely adopted, expect numerous policy changes specifically about "AI plagiarism" to impact art and research).
Note there is currently a known issue with fake citations (and having issues with not formatting citations correctly), including one infamous case where a lawyer used ChatGPT and found out cited cases were fake. As Aaron Welborn points out in the above linked "ChatGPT and Fake Citations":
DO NOT ask ChatGPT for a list of sources on a particular topic! ChatGPT is based on a Large Language Model and does not have the ability to match relevant sources to any given topic. It may do OK with some topics or sources, but it may also fabricate sources that don’t exist.
One specific issue is that ChatGPT may be behind current trends and not have access to important peer reviewed and academic sources. For instance, asking ChatGPT "What are some good recent peer reviewed sources on superconductors?", the response informs you that...
As an AI language model, my responses are generated based on a mixture of licensed data, data created by human trainers, and publicly available data. I have not been trained directly on specific scientific journals or have access to proprietary databases. Therefore, I don't have real-time information on the latest peer-reviewed sources on superconductors beyond my knowledge cutoff in September 2021.
Comparatively, Scopus found thousands of results for articles on that topic since that time period and over 150 had double digits or more of citations (with hundreds of additional being cited and referenced in later papers).
All of this can be mitigated by having a reasonable approach to the materials as presented. Use multiple sources and always evaluate your sources. This is true even with Generative AI, and possibly especially true with the current (2023) generation.
In the general spirit of this guide, I asked ChatGPT, "What are the pros and cons of using ChatGPT in a college setting?" The response, edited for length with additions (marked in italics) makes up the remainder of this section. To see the full transcript, click the link above.
Using ChatGPT in a college setting can have several pros and cons. Let's explore them:
Pros:
Accessibility: ChatGPT can provide students with immediate access to information and support. It can be available 24/7, enabling students to seek assistance outside of regular office hours.
Quick responses: ChatGPT can provide quick responses to students' questions, reducing waiting times and improving efficiency. This can be beneficial for addressing common queries, clarifying course materials, or providing general guidance.
Personalized assistance: ChatGPT can adapt its responses based on individual needs and preferences. Note that the value and impact of this will be highly variable between users and it will be some time to truly understand this portion.
Supplemental learning tool: ChatGPT can serve as a supplemental learning tool, offering additional explanations, examples, or perspectives on course topics.
Scalability: ChatGPT can handle multiple inquiries simultaneously, making it scalable for large student populations.
Cons:
Lack of human interaction: ChatGPT cannot fully replace human interaction, which is an essential aspect of the college experience. Some students may prefer face-to-face interactions for a deeper understanding, emotional support, or building relationships.
Limitations in context comprehension: Although ChatGPT is proficient in understanding and generating text, it may occasionally struggle to grasp context accurately.
Subject expertise limitations: ChatGPT's knowledge is based on pre-existing data and may not be up to date with the latest developments. See above for the 2021 issue when it comes to academic and peer reviewed sources.
Ethical considerations: As an AI language model, ChatGPT can potentially exhibit biases or inadvertently promote misinformation present in its training data. Folks sometimes like to argue against bias in such system but one should really read up on the topic first.
Dependency and overreliance: Students may become overly dependent on ChatGPT, neglecting critical thinking and problem-solving skills. Relying solely on the model for answers could hinder the development of independent research and analytical abilities.
It is essential to strike a balance between leveraging the benefits of ChatGPT and maintaining a human-centric educational approach to maximize its potential in a college setting.
Generally, though, well-said ChatGPT.
