You have /5 articles left.
Sign up for a free account or log in.
In writing this article, I sought the advice of Google Bard, Perplexity and Claude 2. In all of my research using generative AI, I use at least three of the established apps. This enables me to spot any responses that seem too far out of line or not credible. By spreading my research through multiple large language models, I can better ensure that I am not being led astray. Over time, this may not be necessary, but as the apps are being fine-tuned, I feel most comfortable being able to compare results.
Bard uses the PaLM 2 LLM. ChatGPT+ and Perplexity Copilot use versions of GPT4. Claude 2 is powered by Anthropic’s proprietary LLM. Using multiple chat bots with different underlying large language models helps to provide a diverse set of perspectives and responses to the same prompt. It only takes a minute to get a full response (even in the case of Google Bard, which, by default, gives three draft responses to each prompt). Then, using follow-up prompts, you can drill down for clarifications and citations.
(Note: An earlier version of this piece included a section with examples and citations suggested by Google Bard about different applications of GenerativeAI at Stanford University, Carnegie Mellon University, and Georgia Institute of Technology. These examples and citations were misleading and have been removed.)
There are many ways in which we can make effective use of AI every day as higher education faculty and administrators. AI can give us help even with some of the most challenging problems we face. For example, the undergraduate dean at John Jay College, Dara N. Byrne, writes that when she started in her position, the senior graduation rate was 54 percent. She used predictive AI to identify students who were most likely to drop out. In a deep data dive, she teamed with DataKind, a nonprofit using data science and AI, to select students for proactive interventions. This resulted in increasing the senior graduation rate to 86 percent.
The prompt I submitted to each of the three apps was “I am writing an article about productive uses of generative AI. It is for an audience of university deans and faculty. This will be published in Inside Higher Ed. My goal is to give those who have less experience with generative AI a sense of how it might be useful in their everyday work. Give me examples how these higher education workers could use generative AI.”
Following are sample excerpts of some of the results from each of the three apps I consulted.
From a search of Claude 2 on Oct. 29:
Research and writing: Generative AI could help faculty and researchers quickly synthesize literature reviews, generate hypotheses, outline papers and proposals, and draft sections of manuscripts. It could save time and enhance productivity. However, human creativity and critical thinking are still essential.
Administrative tasks: For deans and university leaders, generative AI could help draft communications, generate data insights from reports, compile meeting agendas and minutes, and more. It could save time on logistical tasks to focus leadership efforts elsewhere.
Personalized learning: Generative AI tutors could provide customized explanations, practice, and feedback to students, adapting to their strengths and weaknesses. This could promote self-paced mastery and free up faculty time.
The key for higher education is using generative AI thoughtfully and ethically to augment human efforts. With proper oversight and care, it has much potential to enhance teaching, research, creativity, and productivity. But the human touch remains irreplaceable for true learning, discovery, and decision-making.
From Perplexity on Oct. 29:
Automating Tasks
Automated Grading: generative AI can be used to grade assignments, essays, and exams, freeing up faculty time to focus on other tasks. This can be especially useful for large classes where grading can be time-consuming and tedious.
Chatbots: generative AI–powered chatbots can be used to answer student questions, provide support, and even offer personalized recommendations. This can help reduce the workload of faculty and staff, while also improving the student experience.
Improving Student Outcomes
Personalized Learning: generative AI can be used to create personalized learning experiences for students, based on their individual needs and learning styles. This can help improve student outcomes and engagement.
Early Warning Systems: generative AI can be used to identify students who are at risk of dropping out or failing a course, allowing faculty to intervene early and provide support.
Enhancing Research Data Analysis: generative AI can be used to analyze large datasets, such as those generated by scientific research. This can help researchers identify patterns and insights that might be missed by human analysis.
Natural Language Processing: generative AI can be used to analyze and understand large amounts of text, such as research papers, articles, and books. This can help researchers identify trends and insights that might be missed by human analysis.
As you see, the emphases of the results vary considerably. Of course, different prompts will result in different responses, and different apps will emphasize different interpretations of your prompt. The results to an identical prompt in the same app also will be different if run again just a few minutes after the first one.
The primary key to success in use of generative AI is to develop the prompt carefully. However, the second key to success is to use more than one generative AI app. A third key to success is to use generative AI often, daily if possible. Through repetition, you will refine your prompt writing and determine which app is likely to give you the response you seek for the questions and tasks you have.