There has been a remarkable surge in Artificial Intelligence (AI) usage, generating a significant amount of news and interest. The utilization of ChatGPT has skyrocketed, and new tools emerge regularly, aiding in the creation of images, videos, music, and short films. Microsoft has introduced their co-pilots to assist with code generation, writing Word documents, creating PowerPoint presentations based on Word documents, and analyzing data in Excel using simple word prompts. The future appears both exciting and intimidating.
Now, we are exploring the application of AI to support cost and Academic Program management in Higher Education, leveraging our advantage of possessing a vast and detailed data spanning over 15 years across multiple institutions. Our estimation suggests that we have over 1 Terabyte of data at our disposal for training an internal AI tool. This dataset encompasses various information, including the details of each course/subject instance such as when, where, and how it was taught (e.g., Accounting 101, Semester 1, Campus A, classroom). Additionally, it incorporates resource information from finance, HR, payroll, facilities, as well as teaching and student data from timetabling and student management systems. To provide you with an idea of the sheer volume of data points, a previous analysis project we conducted consolidated one year of data from 11 institutions, yielding over 30,000 individual courses/subjects, each with multiple associated costs and metrics. It is important to emphasize that we have been utilizing consolidated and anonymized data for our clients for a number of years now, with their explicit permission. We firmly believe that since it is their data, we should maximize its utilization to support their institutions.
We have already embarked on exploring the best ways to leverage this substantial amount of data to assist our client base. However, it should also be noted that this data can only be used for our clients’ purposes and not for public use, unless our clients authorize the publication of certain portions for the benefit of higher education as a whole, as we did with the University of Melbourne Center for the Study of Higher Education paper.
Recently, at the June CAM-I conference, Pei-yu (Sharon) Chen, the Earl and Gladys Davis Distinguished Research Professor in Business, Chair of the Department of Information Systems, and Director of the AI and Data Analytics (AIDA) Initiative at the W.P. Carey School of Business, Arizona State University, delivered a timely presentation. The presentation focused on Digital Transformation and the move toward AI-powered organizations. Two key points stood out from her presentation.
Firstly, AI is not a final product or solution in itself; instead, an AI-powered organization aims to develop the capability to sense, analyze, and respond swiftly and accurately to changing conditions.
Secondly, to establish a robust and successful AI strategy, it is crucial to have not only big data but also diversity in data and a cross-functional team involved from the beginning.
This is where our consolidated dataset holds immense potential. While individual universities can certainly consolidate their own data over several years to create a large dataset for analysis and AI training, it represents only one perspective. Conversely, a consolidated dataset from multiple institutions can unlock insights that were previously unattainable.
Academic Program Management
When we inquired why our client institutions rely on our model for this purpose, they provided us with the following reasons:
1. Complexity: Academic Program review is a complex task that cannot be effectively performed as an ad-hoc exercise using simplistic cost estimation metrics.
2. Course-Program Association: It necessitates understanding the associations between Courses (Subjects) and Programs to gain comprehensive insights.
3. Interdisciplinary Courses: It involves identifying Courses that span across multiple schools, such as engineering students taking English courses, which adds another layer of complexity.
4. Course-Level Analysis: Teaching is conducted at the Course (Subject) level, making it crucial to accurately capture information such as Academic Full-Time Equivalent (FTE), Delivery Hours, and instructional methods.
5. Student-Course Mapping: Understanding which students are enrolled in which Courses (Subjects) and the linkage to Programs is vital for revenue calculation.
6. Overhead Allocation: Overhead costs are substantial and often increasing in most institutions. Utilizing a simplistic metric like EFTSL or Student Credit Hours for distributing overhead expenses can lead to overestimating costs for large classes and underestimating costs for small classes.
7. Large Data Size: Our models handle extensive data sets, with many exceeding 100 million rows, while conventional tools like Microsoft Excel are limited to around 1 million rows.
8. Data Integration: Academic Program review necessitates data from multiple source systems, introducing a new layer of complexity to the analysis. This data is often stored in separate systems with minimal interconnection. To conduct a thorough analysis, data from Finance, HR, Payroll, Student Management Systems, Timetabling systems, and occasionally Facility Management systems must be gathered, appropriately prepared, cleansed, and loaded into the model. It must then be allocated to the relevant areas within the model, down to the specific Course, along with information about when, where, and how it’s taught.
In fact, the University of Wollongong, one of our client institutions, developed a tool leveraging our model to assist Academic Leadership in planning new Academic Programs. This tool provides them with an immediate five-year financial forecast based on their projected number of students and the required creation of new Courses/Subjects/Units.
Uses of AI
One of our client institutions recently inquired about the potential use of AI to enhance their modeling efforts, and we are pleased to inform you that we are actively pursuing this avenue. While it will require some dedicated research and development on our part, our initial focus will be on leveraging AI to further support Academic Program Management, which remains the core application of our model.
We are fortunate to have access to extensive and detailed datasets from within organizations, as well as consolidated and anonymized datasets from various institutions. By combining these datasets, we aim to offer more than just financial performance forecasting. Our goal is to provide valuable recommendations for Academic Programs that align with the unique needs of individual organizations. These recommendations could take into account factors such as academic staffing, student demographics, quality of education metrics, and more.
We are thrilled about the possibilities this presents and eager to begin exploring the available options. Engaging in discussions with our client institutions is crucial to understanding their specific objectives and identifying the tailored solutions that will help them achieve their goals effectively.
Now to expand on my thinking, what better way to brainstorm how AI can support Higher Ed Academic Program management, than to ask ChatGPT!
PROMPT:
Consider you are the Provost or Dean at a university and you have access to a large amount of data relating to Academic Programs including full cost, revenue, margins, academic FTE, non-academic FTE, student numbers but also a large dataset of anonymized data from multiple other universities with the same metrics. How could AI be used to support Academic Program Management?
RESPONSE:
As the Provost or Dean at a university, having access to a vast amount of data related to Academic Programs and a comprehensive dataset of anonymized information from multiple other universities opens up numerous possibilities for leveraging AI to support Academic Program Management. Here are a few potential applications:
1. Financial Forecasting and Optimization: AI can analyze historical data on costs, revenues, and margins to develop accurate financial forecasts for existing and proposed Academic Programs. By incorporating machine learning algorithms, the models can continuously improve and adjust predictions based on real-time data. This can aid in optimizing resource allocation, identifying cost-saving opportunities, and improving overall financial management.
2. Program Suitability and Recommendations: By leveraging the anonymized dataset from other universities, AI can analyze the characteristics and performance of similar Academic Programs across different institutions. This analysis can help in recommending suitable programs for the university based on factors such as academic staffing, student demographics, success rates, and quality of education metrics. It enables informed decision-making when introducing new programs or modifying existing ones.
3. Student Success and Retention: AI algorithms can analyze data on student demographics, academic performance, engagement levels, and other relevant factors to identify patterns and indicators of student success or at-risk behavior. This information can help Academic Program Management teams develop targeted interventions and support strategies to improve student outcomes, enhance retention rates, and foster student success.
4. Resource Allocation and Optimization: AI can analyze data on academic and non-academic FTE, student numbers, and other relevant metrics to optimize resource allocation within Academic Programs. By identifying areas of inefficiency or underutilization, AI can provide insights on optimizing staffing, course offerings, scheduling, and infrastructure allocation. This can lead to improved program efficiency, reduced costs, and enhanced utilization of resources.
5. Performance Benchmarking: AI can compare the performance of Academic Programs within the university to similar programs in other institutions. By utilizing the anonymized dataset, AI algorithms can identify best practices, areas of improvement, and performance gaps. This benchmarking can assist in setting realistic goals, implementing targeted improvements, and enhancing the overall quality and competitiveness of the Academic Programs.
6. Predictive Analytics for Enrollment and Demand: AI models can analyze historical enrollment data, student preferences, market trends, and other relevant factors to predict future enrollment patterns and program demand. These predictions can guide strategic decision-making regarding program expansion, development of new courses, and resource allocation to meet anticipated demand.
7. Decision Support Systems: AI-powered decision support systems can provide real-time insights and recommendations to Academic Program Management teams. These systems can assist in scenario analysis, what-if simulations, and impact assessments of proposed changes or new initiatives. By leveraging AI, data-driven decision-making can be facilitated, leading to informed and evidence-based choices.
It is important to note that implementing AI in Academic Program Management requires thoughtful consideration of ethical and privacy concerns, as well as close collaboration with relevant stakeholders, including faculty, staff, and students.
This is an interesting list of suggestions, a couple of things I disagree with – one in particular is “the models can continuously improve and adjust predictions based on real-time data.” – based on our experience with the quality and timeliness of university datasets I doubt we will be using “real time” data any day soon, we can make this an aspirational goal!
Student Success and Retention – the models we build for client institutions don’t go to the level of individual student or individual student performance, so this solution can be left to those organizations and products that specifically focus on this.
But in general, the other suggestions are ultimately achievable, and we are excited about the future prospects of using these new AI technologies to support Higher Education management.