Not All Good, Not All Bad, Not All Equal

By Dr. Peg Hohensee

With all the chatter about students using artificial intelligence (AI) in the classroom to complete their assignments, educators should be looking at ways to help students use AI appropriately. This post provides information about the author’s top five AI applications for working with students, along with the advantages and disadvantages of each. The following is intended as an introduction to AI use in higher education, rather than an authoritative work on the subject.

According to a study by Valdiviezo Sir et al. (2025), AI is improving student performance. Specifically, the authors examined more than 20 primary studies and determined that a key impact occurs when using AI to tailor content to students at public institutions. However, faculty members may not be receiving the professional development needed to guide students in the proper use of AI, and faculty may see AI applications as a threat to academic integrity (Avello & Aranguren Zurita, 2025).

In my experience, many college and university educators are opposed to AI use, but as a mathematician, I think it should be compared to the use of calculators. In high school, I was required to learn to complete trigonometric calculations using a slide rule, even though calculators could make the same calculations faster and more accurately. Looking back, I see that as a missed opportunity to learn how to effectively use a calculator, and more importantly, to develop a deeper understanding of trigonometry, which I dislike to this day.

As educators, we can teach students how to effectively and ethically use AI in order to help them understand that the software should be used to go beyond: do more, understand more, explore more, learn more. I believe we should also teach students that different AI platforms have different purposes and that it is important to select the one that best meets the students’ specific needs. But this will not happen in a vacuum. Avello and Aranguren Zurita (2025) recommend on-going faculty development in promoting responsible use of AI by students and the function of the AI applications.

Some Differences in AI Platforms

At the time of this writing, a quick Google search indicates there are over 70,000 AI apps. Not all AI programs are created equal, and none of us has the time to personally explore in detail the advantages and disadvantages of each of these programs. Therefore, I want to create an entry point for those looking to explore applications that are relatively easy to use for students and faculty. What follows is based on my own personal experience with these specific AI platforms and is not intended to be the ultimate source of information on these applications or all things artificial intelligence. Instead, this list presents my five favorite AI applications, in order of my preference.

1. Perplexity

Advantages. Perplexity is a generative AI that can create summaries and answers to specific questions. My favorite AI, Perplexity provides better sources (with links to the sources) than ChatGPT or Gemini, especially when you set the source options to “academic” in the search feature. Great for research, Perplexity will crawl the web, identifying and summarizing information. Users can then follow the links to and use the original source material, which provides a better reference for academic work. Recently, Perplexity has added Study Mode (Perplexity, 2025), which addresses some of the items that I would have previously listed as disadvantages. Study Mode allows students to create flashcards, includes interactive quizzes, and provides step-by-step explanations (Perplexity, 2025).

Disadvantages. At this time, Perplexity will not create mind maps or outlines, and many of the items created are not editable, which may significantly limit use outside of research and basic study materials. Additionally, Perplexity will not crawl or transcribe audio files, requiring that audio files be first transcribed and uploaded before they can be summarized.

• Gemini

Advantages. Backed by the power of Google, Gemini is also a generative AI capable of finding information and creating multiple study materials, including summaries, study guides, practice quizzes, flashcards, study plans, podcasts/audio overviews, and more (Ayyagari, 2024). Gemini also claims nearly constant updates, making for a very dynamic user experience (Gemini, 2025).

Disadvantages. Gemini will often return information that is overly generic or largely irrelevant. With all AI prompts, the narrower and more focused the question, the more likely the AI will provide usable and relevant information. This seems to be particularly true with Gemini.

3. Notebook LM

Advantages. Unlike the other applications in this list, Notebook is not a generative AI and will not crawl the web for sources. Notebook LM will summarize and synthesize only the material provided, making it a great option when faculty or students want to limit the material to what is presented in the course. Notebook LM can be used to create the same types of study materials as Gemini since it is also a Google product.

Disadvantages. Because Notebook LM is not a web crawler, the user must provide all the information to be used by the AI in either text or audio format. Notebook LM will not find additional resources related to question prompts and is therefore not a good option for initial research.

4. ChatGPT

Advantages. Arguably one of the most well-known AI applications, ChatGPT is an easy-to-use generative AI that may already be very familiar to students and faculty. ChatGPT provides an “entry-level” experience to AI because responses are concise and include far less background information than Gemini or Perplexity, making it easier for students to consume.

Disadvantages. In my opinion, the biggest disadvantage of ChatGPT is that it often returns information that includes sources that are not academically appropriate or reliable. For example, I have received returns from question prompts with material from Wikipedia and Reddit. Additionally, many of the sources for the reply remain hidden unless a specific request for the sources is included in the prompt, so students may be unaware that the material is academically inappropriate to cite.

5. Claude

Advantages. An excellent option for users who want to work with computer code or complete an analysis, Claude is a generative AI that appears to focus more on this type of information than other AI platforms (Anthropic, 2024). Claude can generate both quantitative and qualitative analyses, including graphs and charts, and provides a code window during the analysis that allows users to review the code being created and processed in real time.

Disadvantages. Information provided by Claude is often dense and difficult to understand when compared to that created by other generative AI applications. Additionally, Claude’s graphic output is not editable.

Conclusions

I don’t think we have reached the point (nor do I hope we ever do) when faculty can abdicate their teaching duties or students relinquish their responsibility for learning. AI, much like the calculator, can be seen as a tool to replace many of the menial responsibilities of research and materials creation in order to free up time to explore the content more thoroughly. Teaching students to effectively use AI platforms extends beyond selecting applications. Students should also be taught that cutting/pasting directly from an AI isn’t appropriate, and that it is important to evaluate the source(s) of information provided by the AI tool. AI cannot replace the human brain in terms of critical thinking and synthesis, but creating a specific prompt for an AI and then appropriately using the information provided by that tool can save time and frustration.

References

Anthropic. (2024, March 4). Introducing the next generation of Claude. https://www.anthropic.com/news/claude-3-family

Avello, D., & Aranguren Zurita, S. (2025). Exploring the nexus of academic integrity and artificial intelligence in higher education: A bibliometric analysis. International Journal for Educational Integrity, 21(1), 1–14. https://doi.org/10.1007/s40979-025-00199-2

Ayyagari, K.C. (2024, December 6). Build and refine your audio generation end-to-end with Gemini 1.5 Pro. AI & Machine Learning. https://cloud.google.com/blog/products/ai-machine-learning/learn-how-to-build-a-podcast-with-gemini-1-5-pro

Gemini. (2025). Gemini drops. https://gemini.google/gemini-drops/#:~:text=Gemini%20is%20constantly%20evolving%2C%20but%20Gemini%20Drops,Gemini%20to%20create%2C%20research%2C%20and%20do%20more

Perplexity (2025). What is study mode? https://www.perplexity.ai/help-center/en/articles/12120542-what-is-study-mode

Valdiviezo Sir, V.M., Reátegui, J.A., Guevara, R.C., Revilla, A.C., & Valqui, C.V. (2025). Artificial intelligence and educational quality in higher education: A systematic review. TEM Journal, 14(4), 3794–3802. https://doi.org/10.18421/TEM144-82

Biographical Note: Dr. Peggy Hohensee is the Mathematics Chair for Purdue Global (PG). She has 25+ years of experience in higher education and holds an MEd in curriculum and instruction, an MS in applied mathematics, an MBA, and a PhD in educational leadership and research from Louisiana State University.