Prologue and epilogue by Isabel Fischer (Warwick Business School)
Since the creation of our task and finish groups on evaluating and shaping the opportunities of AI in Higher Education (for further information see here: https://blogs.warwick.ac.uk/wjett/entry/ai_for_good/) we have had many very insightful ‘human to human’ discussions. The aim of this blog is, however, to show what ‘machines’ draw and write on the topic, to illustrate both the strengths and the weaknesses of generative AI.
The images at the start of the blog represent DALL.E’s take on the topic. Having always been untalented in drawing, I use DALL.E in my business teaching to show live how AI translates the topic we discuss by imitating distinct artists. Students enjoy how DALL.E (OpenAI) shows other – very creative - images while we wait for AI ‘to think & draw’ our images – there are always four images coming up for each prompt. Of course, we also discuss the risks, threats, and ethics behind AI (here an open-access case study which we wrote on AI ethics in the field of ESG investing: https://journals.sagepub.com/doi/full/10.1177/2043886920961782).
While DALL.E generates images that I could not have drawn, ChatGPT generates text that I could have written better and more concisely, with different logical priorities. I could - and actually did - create visualisations to illustrate my points. Nevertheless, ChatGPT did beat me by far on ‘speed’. For anybody who has not yet asked ChatGPT any questions, here are the ChatGPT responses based on five prompts:
- Opportunities for AI in HE
- Strategy for implementing and integrating AI in HE
- Developing research on the opportunities of AI in HE
- Developing an AI based marking and feedback tool
- Common areas of weakness of students’ academic writing
Opportunities for AI in HE
Artificial intelligence (AI) is transforming the world around us, and higher education (HE) is no exception. The potential opportunities of AI in HE are vast, ranging from improving learning outcomes and enhancing student engagement to streamlining administrative tasks and reducing costs. In this article, we explore some of the key opportunities of AI in HE and how they can be realized.
- Personalized learning: AI can help personalize the learning experience by adapting the content, pace, and style of instruction to the individual needs and preferences of each student. This can be achieved through intelligent tutoring systems, which use machine learning algorithms to analyze student performance data and provide customized feedback and support. Personalized learning can improve student engagement, increase retention rates, and enhance learning outcomes.
- Predictive analytics: AI can be used to analyze large datasets of student information, such as grades, attendance records, and behavioral data, to predict future performance and identify at-risk students who may need additional support. This can help educators intervene early and provide targeted interventions to improve student success.
- Virtual assistants: AI-powered virtual assistants can help students with routine administrative tasks such as scheduling, registration, and accessing resources. This can free up staff time and reduce wait times for students, improving their overall experience.
- Automated grading: AI can be used to automatically grade assignments and provide feedback to students, reducing the workload on instructors and improving the consistency and objectivity of grading. This can also provide students with immediate feedback, enabling them to improve their performance more quickly.
- Research and data analysis: AI can be used to analyze large datasets of research publications and help identify trends, patterns, and gaps in the literature. This can help researchers stay up-to-date with the latest developments in their field and generate new insights.
- Adaptive assessments: AI can be used to create adaptive assessments that adjust the difficulty and content of questions based on the student's level of knowledge and ability. This can help educators more accurately assess student performance and provide targeted feedback and support.
- Smart campus: AI can be used to optimize campus operations, such as energy management, facility maintenance, and security. This can reduce costs and improve the overall efficiency of campus operations.
While the opportunities of AI in HE are significant, there are also challenges that need to be addressed, such as ethical considerations, data privacy concerns, and ensuring that AI is used to enhance, rather than replace, human interaction and support. To fully realize the potential of AI in HE, it is essential that educators, researchers, and policymakers work together to develop responsible and effective strategies for implementing and integrating AI technologies.
Strategy for implementing and integrating AI in HE
Implementing and integrating AI technologies in higher education (HE) requires careful planning and consideration to ensure that these technologies are used in a responsible and effective manner. Here are some responsible and effective strategies that HE institutions can use to implement and integrate AI technologies:
- Develop a clear vision and strategy: HE institutions should develop a clear vision and strategy for how AI technologies will be used to enhance the student experience, improve learning outcomes, and support research. This should involve identifying specific use cases and goals for AI, as well as assessing the potential benefits and risks.
- Foster collaboration and communication: Implementing and integrating AI technologies requires collaboration and communication across departments and stakeholders. HE institutions should establish cross-functional teams that include educators, administrators, IT professionals, and students to ensure that all perspectives are considered and to foster buy-in and support for AI initiatives.
- Ensure transparency and ethical considerations: HE institutions should ensure that the use of AI technologies is transparent and ethical. This includes being clear about how AI technologies are being used, ensuring data privacy and security, and ensuring that AI is used to enhance, rather than replace, human interaction and support.
- Invest in staff training and development: AI technologies require specialized skills and knowledge to implement and use effectively. HE institutions should invest in staff training and development programs to ensure that educators and administrators have the necessary skills and knowledge to use AI technologies in their work.
- Pilot and evaluate AI initiatives: Before scaling up the use of AI technologies, HE institutions should pilot and evaluate their effectiveness. This involves testing AI technologies in small-scale pilots, collecting feedback from stakeholders, and using data to evaluate their impact on learning outcomes, student engagement, and other key metrics.
- Engage with the wider AI community: HE institutions should engage with the wider AI community to stay up-to-date with the latest developments and trends in AI. This includes attending conferences and workshops, collaborating with other institutions, and partnering with industry experts and researchers.
- Align AI initiatives with institutional goals and values: Finally, HE institutions should ensure that their AI initiatives are aligned with their institutional goals and values. This involves considering the unique needs and priorities of the institution, as well as the needs and preferences of students, educators, and other stakeholders. By aligning AI initiatives with institutional goals and values, HE institutions can ensure that they are using these technologies in a responsible and effective manner.
Developing research on the opportunities of AI in HE
There are several ways to develop ideas for future opportunities of AI in higher education and organize it as a research topic:
- Literature review: Conducting a literature review of existing research on AI in higher education can help identify the gaps and potential areas for future research. This can involve searching for relevant articles, books, and reports in databases such as Google Scholar, ACM Digital Library, and IEEE Xplore.
- Expert interviews: Talking to experts in the field, such as educators, researchers, and AI professionals, can help identify emerging trends and potential opportunities for AI in higher education.
- Case studies: Conducting case studies of AI implementation in higher education institutions can provide insights into the benefits and challenges of adopting AI technologies.
- Surveys: Conducting surveys of stakeholders such as students, faculty, and administrators can help identify their needs and preferences for AI in higher education.
- Workshops and conferences: Attending workshops and conferences on AI in higher education can provide opportunities to learn from experts and exchange ideas with peers.
Once you have developed ideas for future opportunities of AI in higher education, organizing it as a research topic can involve the following steps:
- Identifying research questions: Based on the ideas developed, identify research questions that are relevant, feasible, and significant.
- Defining the scope of the research: Clarify the boundaries of the research topic and define the key concepts and variables.
- Developing a research methodology: Decide on the appropriate research methods, such as qualitative, quantitative, or mixed methods, and develop a research design.
- Conducting data collection: Collect data using appropriate methods such as surveys, interviews, and case studies.
- Analyzing data: Analyze the data using appropriate statistical or qualitative analysis techniques.
- Drawing conclusions and making recommendations: Based on the findings, draw conclusions and make recommendations for future research and practice.
- Writing up the research: Write up the research in a clear and concise manner, following the conventions of academic writing.
When reviewing emerging trends of AI in higher education after conducting qualitative research, a theory that could be useful is grounded theory.
Grounded theory is a research methodology that involves systematically collecting and analyzing qualitative data to develop a theory or explanation of a phenomenon. The approach emphasizes the need to generate theory from the data itself, rather than imposing preconceived notions or theories on the data.
When conducting a qualitative study of emerging trends of AI in higher education, the grounded theory approach can help identify key themes and patterns in the data that may point towards emerging trends. These trends can then be used to develop a theory that explains how AI is being used in higher education, what the benefits and challenges are, and what the future implications may be.
Grounded theory can also be helpful in identifying gaps in the literature and areas for future research. By systematically analyzing the data, researchers can identify areas where more research is needed, and use the emerging trends to develop research questions that are relevant and meaningful.
Developing an AI based marking and feedback tool
Designing an AI system that can be used to automatically grade assignments and provide personalized feedback to students can be a complex task, but it can be accomplished by following these steps:
- Define the grading criteria: The first step is to define the grading criteria for the assignment. This involves breaking down the assignment into its key components and determining how each component will be evaluated. The grading criteria should be clearly defined and objective, so that the AI system can easily understand and apply them.
- Create a training dataset: To train the AI system, you need to create a training dataset that includes a set of annotated assignments. These annotated assignments should be graded by human experts according to the defined grading criteria. The dataset should be large enough to ensure that the AI system can learn to recognize a variety of different assignment types and styles.
- Choose the appropriate machine learning algorithm: There are several machine learning algorithms that can be used for automatic grading, including decision trees, support vector machines, and neural networks. The choice of algorithm will depend on the complexity of the grading criteria and the size of the training dataset.
- Train the AI system: Once you have chosen the appropriate machine learning algorithm, you can train the AI system using the annotated training dataset. The system should be trained to recognize patterns and relationships in the data, and to identify which components of the assignments are important for grading.
- Test the AI system: After training the AI system, you should test it using a set of unseen assignments that have not been included in the training dataset. This will help you evaluate the system's accuracy and identify any areas where it needs further improvement.
- Provide personalized feedback: To provide personalized feedback to students, you can use the AI system to identify areas where a student may need additional support or guidance. For example, if a student consistently struggles with a particular component of the assignment, the system can provide targeted feedback and resources to help the student improve.
- Continuously improve the AI system: Finally, you should continuously monitor and improve the AI system. This involves regularly updating the training dataset, evaluating the system's performance, and incorporating feedback from students and educators to refine the grading criteria and feedback mechanisms.
In summary, designing an AI system for automatic grading and personalized feedback involves defining the grading criteria, creating a training dataset, choosing the appropriate machine learning algorithm, training the system, testing it, providing personalized feedback, and continuously improving the system. By following these steps, you can develop an effective and efficient AI system that enhances the learning experience for students and educators alike.
Common areas of weakness of students’ academic writing
Academic writing is an essential skill for students in higher education. However, there are common areas of weakness that students often struggle with. Here are some of the most common areas of weaknesses for students' academic writing in higher education:
- Poor organization and structure: Students often struggle with organizing their ideas and structuring their writing in a way that is clear and logical. This can make their writing difficult to follow and reduce the overall effectiveness of their argument.
- Weak thesis statements: A strong thesis statement is essential for guiding the direction of an academic paper. However, many students struggle to develop a clear and concise thesis statement that effectively summarizes the main argument of their paper.
- Inadequate research and citation skills: Effective academic writing requires the use of credible sources to support arguments. However, many students struggle with finding appropriate sources, evaluating their credibility, and correctly citing them in their writing.
- Poor grammar and syntax: Grammatical errors and poor syntax can make writing difficult to read and understand. This can detract from the clarity and effectiveness of students' arguments.
- Lack of clarity and concision: Academic writing should be clear and concise, but many students struggle with using language that is overly complex or wordy. This can make their writing difficult to read and reduce the impact of their argument.
- Inappropriate tone or style: Students may struggle to adopt an appropriate tone or style for academic writing, which can lead to writing that is overly informal or lacking in scholarly rigor.
- Failure to address the assignment prompt: Finally, students may struggle with addressing the assignment prompt in a way that fully meets the expectations of the instructor. This can result in writing that is off-topic or fails to fully address the main requirements of the assignment.
Overall, addressing these common areas of weakness in academic writing can help students improve their writing skills and achieve greater success in their academic pursuits.
Epilogue: For comparison to the above, here are some ChatGPT responses to prompts by Chaoran Yang, a student who is part of my consulting module. I suggested to all students of that cohort to try ChatGPT for idea generation at the start of a consulting exercise for the Warwick Esports centre. Similar to my findings from above, the students also felt that ChatGPT did not provide any major insights. Possibly because we did not ask the right questions, and, more importantly, because ChatGPT is still in its infancy.
In our team we have spent over 2,000 hours developing the Warwick AI Essay Analyst. We used a mixture of non-AI rule-based statistical features and deep-learning algorithms and databases, e.g., Pytorch, Hugging face framework, and Transformer (for further information on our AI-based tool see here: https://nationalcentreforai.jiscinvolve.org/wp/2022/11/16/interested-in-receiving-formative-feedback-on-your-draft-essays-and-dissertations-on-demand-introducing-warwicks-ai-essay-analyst/).
With the current progress in the field of generative AI, developments of future tools will be faster – let’s work together to ensure that all tools, whether developed in-house or bought / endorsed by the university have robust ethical underpinnings. My final suggestions for readers is to review here the Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators, produced by the Office of the European Union: https://data.europa.eu/doi/10.2766/153756