ChatGPT launched approximately one year ago this month. Since its availability to the public, the number of tools and apps that utilize the Generative AI technology have exploded. Dozens of faculty members and academic leaders I follow on LinkedIn have posted articles about their utilization of these tools and anecdotal evidence of student utilization.
I recently read the publication, GenAI in Higher Education: Fall 2023 Update Time For Class Study. The research, conducted by Tyton Partners and sponsored by Turnitin, expands the findings from surveys of faculty and students reported in the Spring and Fall of 2023. Since I had missed the publication of the Spring 2023 Time for Class Study, I reviewed it first.
Time for Class Study – Spring 2023
The foundation for the Gen AI report is the Time for Class: Bridging Student and Faculty Perspectives on Digital Learning report issued in June 2023. Tyton Partners conducted three surveys in the spring of 2023 receiving feedback from 2,048 students, 1,748 instructors, and 306 higher education administrators.
Barriers to improving the promise of digital learning
Several barriers to improving “the promise” of digital learning were identified in the surveys. These were:
- Lack of reliable access to technology – many students lack access to stable internet, devices, and applications. The problem is acute for community college students and students of color but persists across all institutional sectors and student demographics.
- Misalignment of instructor and faculty preferences – students strongly prefer hybrid and digital modes of instruction, and instructors prefer face-to-face instruction. The same preference applies to course materials where students prefer digital, and instructors prefer print.
- Barriers to accessing course materials – students prefer access to digital materials that are available on day one of class and that reduce costs. Faculty are aware of student affordability concerns. Administrators are not always aware of their supply channels’ impact on first day of class materials availability.
- Developing course–specific community – students who begin courses without a set community (first year students and online students) are more likely to use digital tools that increase course engagement including collaboration tools and study aids.
Key instructor challenges in digital learning
Three key instructor challenges were identified in the survey. These were:
- Tool selection is custom for each course – Instructors choose core digital materials from a wide range of options with 25% of faculty using a combination of courseware, e-text, and open education resources across their teaching load. In addition, they supplement digital materials with digital tools to support assessment, proctoring, student collaboration and other class functions.
- AI is here to stay – The release of open-use generative AI writing tools creates a new challenge in preventing student cheating. Faculty and administrators lag students in tool usage. Administrators and faculty must experiment with these tools to develop effective and informed policies and/or integration into teaching and learning.
- Good teaching matters, but institutions need to support it – Students who report that their instructors use more evidence-based teaching practices also report more positive outcomes such as belonging and confidence that they will pass the course.
Spring Survey Data, Charts, and Tables
The findings of the Spring 2023 survey as summarized are more useful when you review the detailed charts and tables that I did not include. I highly recommend downloading a copy of the report, particularly for administrators looking for data to support enhancing their digital teaching capabilities.
Fall Survey Update
Survey results from the Spring survey about the use of Generative AI (“AI is here to stay” bullet above) stimulated a more focused pulse survey for the Fall about the use of Gen AI writing tools and how they are being used by students and instructors.
Chat GPT was introduced to the public in November 2022. By the time of the March 2023 Spring surveys, 9% of faculty were using a Gen AI tool and 27% of students were using a Gen AI tool. As noted in the September 2023 survey, faculty usage of Gen AI writing tools more than doubled to 22% and student usage nearly doubled to 49%. Viewed another way, faculty usage was six months behind students as of September 2023.
Students’ perspectives on the impact of Gen AI on their learning was relatively unchanged from Spring to Fall (approximately 50% positive impact) whereas the faculty negative impact on student learning percentage decreased from 50% to 39%. The faculty decrease was likely due to more faculty finding the time to utilize a Gen AI writing tool.
Will Gen AI writing tools be used for work?
One of the questions built specifically for the Fall survey was whether it was believed that students would need to know how to effectively use generative AI writing tools to succeed in professional settings. Approximately 55% of students who used AI believe they will need to know how to effectively use Gen AI writing tools.
Perhaps it’s the perceived difference between higher ed and the workforce, but 75% of instructors surveyed who use AI believe that students will need to know how to effectively use Gen AI writing tools to succeed in professional settings. Only 8% of instructors believe students will not need to know how to effectively use Gen AI writing tools.
Will Students Continue to Use Gen AI Tools if Banned?
In the Spring survey, approximately 75% of students using AI indicated that they would be extremely likely, likely, or somewhat likely to continue to use Gen AI writing tools even if their instructor or school banned them. Those percentages were unchanged among students surveyed in the Fall survey.
How do Faculty Use Gen AI Writing Tools?
The top reason cited by faculty using Gen AI writing tools is to run prompts through an AI tool to see what students see (43%). The second highest reason is to teach students how to make effective use of generative AI writing tools (35%). The third highest reason is to create more engaging in-class activities (29%). Generating writing assignments was ranked fourth at 26%.
Faculty Permitted Use Cases for Gen AI Writing Tools
Faculty were asked to “indicate which of the below you would allow students to use generative AI writing tools for in your courses. Select all that apply.” There were three selections that more than 50% of the faculty indicated they would allow.
The three selections were: brainstorming ideas (73% for AI users), outlining structure (59% for AI users), and editing assignment (52% for AI users). These percentages were similar to students’ AI use cases.
There were two selections where the faculty’s willingness to allow usage by students varied significantly from the students’ rankings. These were: to write small parts of an assignment (faculty 21% vs. students 45%) and to write large parts of an assignment (faculty 6% vs. students 27%).
The Tyton researchers note that 45% of faculty use essays as an assessment. It’s no surprise that faculty are much less willing to allow students to use generative AI writing tools for writing small parts or large parts of an assignment. Faculty were also queried on what percentage of an essay flagged as being AI-generated content would cast doubts on the essay’s integrity. The average threshold was 35% for faculty AI users and 29% for faculty non-users.
Trends to Watch
The Tyton researchers provided a list of trends that faculty and administrators should watch. These are:
- Students and faculty believe Gen AI has the potential to improve learning outcomes.
- Gen AI has the potential to become ubiquitous for all actors in higher education.
- Gen AI can increase productivity and efficiency for instructors and students.
- Gen AI brings ethical considerations and challenges to equity in education.
A Few Observations
Tyton Partners has significant experience surveying students and instructors about digital teaching and learning. The demographics and distribution of student and instructor survey participants appear to mirror U.S. higher education in general.
Tyton’s researchers recommend that administrators and faculty receive instruction in the utilization of Gen AI writing tools. They strongly believe that familiarity with these Gen AI writing tools will help instructors and administrators recognize broader applications for the use of these tools as well as build useful policies for how to use these tools.
Some of the issues related to access to Gen AI writing tools may be reduced when applications like Microsoft Office and Google Docs incorporate them. I suspect that will happen within the next six to 12 months. Many colleges provide low-cost or no-cost access to these productivity applications.
Continued development of add-on applications for Gen AI tools has resulted in the creation of guard rails that minimize the hallucination of responses from the earlier and more generic models. Between the demand from institutions as well as the competition for market share, I believe learning to utilize these tools will be enhanced.
I know at least one startup, Flint, that provides a generative AI platform for K -12 schools to use that does not train its models on students’ prompts and allows tight control (invitation only and the ability to review student usage) for administrators and faculty review. I am sure that there will be more.
Lastly, if I were an administrator trying to find the “perfect” training class for my faculty and administrators to learn to use Gen AI, I wouldn’t spend too much time searching. Ethan Mollick’s recent blog post recommends that just using AI will teach you to use AI. Set an example and share your rigor of using Chat GPT4 or another tool once a day.
Buy a group license from Open AI or reimburse your faculty and staff for subscribing. Set up a Slack account or some other means so that everyone can share ways in which they have utilized the tool for the benefit of their class, their work, or the school. Don’t wait for the “best practices” books to be published before you start. By then, you’ll be even further behind.