Chapter 11: AI in Higher Education and Interdisciplinary Learning |
A Brief Summary at the Outset |
This chapter examines how generative artificial intelligence tools are being used in higher education, with a particular focus on computer science, electrical engineering, and economics. Instructors in these fields have adopted tools such as Perplexity and Google NotebookLM to customize learning experiences, design dynamic classroom activities, and create realistic scenarios that connect theory with practice. Survey-based evaluations show that student engagement has increased, but acceptance levels vary widely. Concerns remain about how quickly AI should be integrated into courses and how to balance AI support with instructor-led guidance. The pages that follow describe concrete applications across multiple industries and disciplines, compare tools, and offer a detailed summary at the end. |
|
1. Introduction: Why Higher Education Is a Special Case |
Higher education sits at a crossroads between pure research and practical workforce preparation. Unlike primary or secondary education, where curricula are relatively standardized, universities and colleges serve students who are preparing for specialized careers in engineering, medicine, business, law, and the humanities. This diversity means that a single AI tool cannot serve everyone equally well. |
At the same time, higher education faces pressure from several directions. Class sizes are growing in many introductory courses. Students arrive with uneven preparation. Employers expect graduates to be familiar with modern digital tools. And instructors must often balance teaching with research and administrative duties. Generative AI offers a way to address some of these pressures, but it also raises questions about academic integrity, dependency, and the purpose of education itself. |
The three fields highlighted in this chapter, computer science, electrical engineering, and economics, illustrate these tensions clearly. Computer science students often learn to build AI systems, so they must understand both the capabilities and the limits of generative models. Electrical engineering students work with hardware and signals, where AI can help simulate complex systems but cannot replace hands-on laboratory experience. Economics students study human behavior, markets, and policy, where AI can generate realistic scenarios but must be used carefully to avoid reinforcing unrealistic assumptions. |
|
2. The Tools in Brief: Perplexity and Google NotebookLM |
Before describing specific applications, it helps to understand what these tools do at a basic level. |
Perplexity is an AI-powered search and answer engine. A user asks a question in natural language, and Perplexity responds with a synthesized answer that includes citations to web sources. Unlike a traditional search engine that returns a list of links, Perplexity tries to provide a direct, readable answer while still showing where the information came from. This makes it useful for students who need to quickly grasp a new topic or verify facts. |
Google NotebookLM is a tool for working with documents. A user uploads sources such as lecture notes, research papers, or textbook chapters. NotebookLM then allows the user to ask questions about those sources, generate summaries, create study guides, and even produce audio overviews that sound like a podcast discussion between two hosts. The key difference from Perplexity is that NotebookLM is grounded in the user's own uploaded materials, which reduces the risk of hallucinated facts. |
Other generative AI tools exist, including general-purpose chatbots and specialized tutoring systems. But Perplexity and NotebookLM are particularly interesting for higher education because they emphasize citations, source grounding, and document analysis, all of which align with academic values. |
|
3. Computer Science: Teaching Students to Build and Critically Evaluate AI |
Computer science instructors face a unique challenge. Their students will likely build AI systems after graduation. Therefore, these students need to understand not only how to use AI tools but also how those tools work, where they fail, and what ethical questions they raise. |
3.1 Customizing Learning Paths |
In a typical introductory programming course, students have widely different backgrounds. Some have been coding since middle school. Others are writing their first line of code. Perplexity can help by generating personalized explanations. For example, a student who struggles with recursion can ask Perplexity to explain it using a simple analogy. A more advanced student can ask for a comparison of recursion and iteration in terms of memory usage. The instructor does not have to prepare separate materials for every level. |
NotebookLM takes this further. An instructor can upload a set of programming assignments, sample solutions, and common error messages. Students can then ask NotebookLM why a particular error occurs and how to fix it. Because NotebookLM is grounded in the uploaded materials, the answers tend to be consistent with the course's own conventions. |
3.2 Dynamic In-Class Activities |
Lectures can become stale if the instructor simply reads from slides. Generative AI can help design interactive activities on the fly. For instance, an instructor teaching data structures might ask Perplexity to generate a set of real-world scenarios where a stack would be more appropriate than a queue. The instructor can then present these scenarios to the class and ask students to justify their choices. |
In a software engineering course, students might use NotebookLM to analyze a set of user stories and generate test cases. The instructor can upload a project description and ask students to work in groups to identify edge cases. NotebookLM can then be used to check whether the groups missed any important scenarios. |
3.3 Realistic Scenarios Bridging Theory and Practice |
Computer science theory can feel abstract. Generative AI can create realistic scenarios that make theory concrete. For example, in a course on algorithms, students might learn about shortest-path algorithms. Perplexity can generate a scenario involving a delivery company that needs to optimize routes across a city. Students then apply the algorithm to the scenario and discuss practical constraints such as traffic and fuel costs. |
In a machine learning course, students might study bias in training data. NotebookLM can help them analyze a dataset description and identify potential sources of bias. The instructor can upload news articles about AI failures and ask students to trace those failures back to design decisions. |
3.4 Survey Results and Concerns |
Surveys of computer science students using these tools show increased engagement. Students appreciate the ability to get immediate answers and to explore topics at their own pace. However, acceptance is not universal. Some students worry that they are becoming too dependent on AI and losing the ability to debug code on their own. Others feel that AI answers are sometimes too confident or too vague. |
Instructors share these concerns. They worry about the optimal pacing of AI integration. If AI is introduced too early, students may never develop fundamental skills. If it is introduced too late, students may feel that the course is disconnected from modern practice. There is also the question of balance. How much should the instructor lead, and how much should the AI supportToo much AI can make the instructor feel redundant. Too little AI can make the course feel outdated. |
|
4. Electrical Engineering: Simulation, Design, and Laboratory Work |
Electrical engineering is a hands-on field. Students work with circuits, signals, power systems, and communication devices. Generative AI cannot replace a soldering iron or an oscilloscope. But it can enhance the learning experience in several ways. |
4.1 Customizing Learning Paths |
Electrical engineering students often struggle with mathematics. Differential equations, complex numbers, and Fourier transforms can be daunting. Perplexity can generate step-by-step explanations tailored to a student's current level. For example, a student who does not understand phasors can ask for a geometric interpretation. Another student who wants to understand the connection to differential equations can ask for that instead. |
NotebookLM can be used to create a personalized study guide from lecture notes and textbook chapters. The student uploads the materials and asks NotebookLM to generate a list of key concepts, practice problems, and common pitfalls. This is especially useful before exams. |
4.2 Dynamic In-Class Activities |
In a circuits course, the instructor might ask Perplexity to generate a set of design challenges. For example, 'Design a voltage divider that produces 3.3 volts from a 5-volt source while drawing less than 1 milliampere.' Students work in groups to solve the challenge. The instructor can then use Perplexity to generate additional constraints, such as 'Now minimize power consumption' or 'Now account for a 10 percent tolerance in resistor values.' |
In a signals and systems course, NotebookLM can be used to analyze real-world signals. The instructor uploads a set of audio recordings or sensor data. Students ask NotebookLM to describe the frequency content and to suggest appropriate filters. This bridges the gap between mathematical theory and practical application. |
4.3 Realistic Scenarios Bridging Theory and Practice |
Electrical engineering projects often involve trade-offs. Generative AI can create realistic scenarios that force students to make decisions. For example, in a power electronics course, students might be asked to design a solar inverter for a rural clinic. Perplexity can generate details about the clinic's energy needs, the local climate, and the budget. Students then must choose between different topologies and justify their choices. |
In a communications course, students might study modulation schemes. NotebookLM can help them analyze a scenario involving a deep-space probe that must transmit data over a noisy channel. The instructor uploads technical papers about error-correcting codes. Students use NotebookLM to compare different codes and recommend one for the mission. |
4.4 Survey Results and Concerns |
Surveys of electrical engineering students show that engagement increases when AI is used for simulation and scenario generation. Students enjoy the ability to explore 'what if' questions without risking real hardware. However, some students feel that AI-generated scenarios are too clean and do not capture the messiness of real laboratory work. Others worry that they are not getting enough hands-on practice. |
Instructors note that AI is best used as a supplement, not a replacement. Laboratory work remains essential. The optimal pacing question is particularly acute in electrical engineering because the curriculum is tightly sequenced. Introducing AI too early might confuse students who have not yet mastered the fundamentals. Introducing it too late might leave them unprepared for industry. |
|
5. Economics: Models, Policy, and Human Behavior |
Economics is a social science. It uses mathematical models to understand human behavior, markets, and policy. Generative AI can help students explore these models, but it must be used carefully because economic models are simplifications of reality. |
5.1 Customizing Learning Paths |
Economics students often struggle with abstract models. Perplexity can generate intuitive explanations. For example, a student who does not understand the concept of elasticity can ask for a real-world example. Another student who wants to understand the mathematical derivation can ask for that instead. This flexibility allows each student to build understanding at their own pace. |
NotebookLM can be used to create a personalized reading guide. The instructor uploads a set of research papers and news articles. Students ask NotebookLM to summarize the main arguments and to identify areas of disagreement. This helps them develop critical thinking skills. |
5.2 Dynamic In-Class Activities |
In a microeconomics course, the instructor might ask Perplexity to generate a set of market scenarios. For example, 'What happens to the price and quantity of coffee if a major producer experiences a drought' Students work in groups to analyze the scenario using supply and demand diagrams. The instructor can then ask Perplexity to add complications, such as 'Now assume that consumers can switch to tea.' |
In a macroeconomics course, NotebookLM can be used to analyze historical episodes. The instructor uploads central bank reports and news articles from a particular period. Students ask NotebookLM to identify the policy decisions and their consequences. This brings history to life. |
5.3 Realistic Scenarios Bridging Theory and Practice |
Economics is often about trade-offs. Generative AI can create realistic scenarios that require students to weigh competing goals. For example, in a public economics course, students might be asked to design a tax policy for a developing country. Perplexity can generate details about the country's economy, its informal sector, and its administrative capacity. Students then must choose between different tax instruments and justify their choices. |
In a behavioral economics course, students might study nudges. NotebookLM can help them analyze a scenario involving a retirement savings program. The instructor uploads research papers about default options and framing effects. Students use NotebookLM to design a nudge and predict its impact. |
5.4 Survey Results and Concerns |
Surveys of economics students show that engagement increases when AI is used to generate realistic scenarios. Students appreciate the ability to apply abstract models to concrete situations. However, some students worry that AI-generated scenarios are too simplistic and do not capture the complexity of real economies. Others worry about the accuracy of AI-generated facts. |
Instructors share these concerns. They note that AI can sometimes generate plausible-sounding but incorrect economic data. This makes it essential for instructors to verify AI outputs before using them in class. The optimal pacing question is also relevant. Economics students need a strong foundation in theory before they can critically evaluate AI-generated scenarios. |
|
6. Cross-Disciplinary Applications: Beyond the Three Fields |
The lessons learned from computer science, electrical engineering, and economics apply to other fields as well. This section briefly describes applications in several other industries. |
6.1 Medicine and Healthcare |
Medical students must learn anatomy, pharmacology, and clinical reasoning. Generative AI can help by generating patient scenarios. For example, Perplexity can create a case of a patient with chest pain, including history, symptoms, and test results. Students then work in groups to develop a differential diagnosis. NotebookLM can be used to analyze clinical guidelines and research papers. Instructors must be careful because medical errors can have serious consequences. AI-generated scenarios must be reviewed by experts. |
6.2 Law and Legal Education |
Law students must learn to read cases, apply statutes, and construct arguments. Generative AI can help by generating hypothetical legal scenarios. For example, Perplexity can create a contract dispute involving a small business. Students then must identify the relevant legal issues and predict how a court might rule. NotebookLM can be used to analyze a set of cases and statutes. Instructors must ensure that AI does not reinforce biases or produce incorrect legal information. |
6.3 Business and Management |
Business students must learn to analyze markets, manage teams, and make decisions under uncertainty. Generative AI can help by generating case studies. For example, Perplexity can create a scenario involving a company that is considering entering a new market. Students then must analyze the risks and opportunities. NotebookLM can be used to analyze financial reports and news articles. Instructors must ensure that AI-generated scenarios are realistic and not overly simplified. |
6.4 Engineering and Design |
Engineering students in fields such as mechanical, civil, and chemical engineering must learn to design systems that meet specifications. Generative AI can help by generating design challenges. For example, Perplexity can create a scenario involving a bridge that must withstand an earthquake. Students then must choose materials and dimensions. NotebookLM can be used to analyze building codes and research papers. Instructors must ensure that AI-generated challenges are safe and realistic. |
6.5 Natural Sciences |
Students in physics, chemistry, and biology must learn to conduct experiments and interpret data. Generative AI can help by generating experimental scenarios. For example, Perplexity can create a scenario involving a chemical reaction that produces an unexpected product. Students then must propose an explanation. NotebookLM can be used to analyze research papers and laboratory manuals. Instructors must ensure that AI-generated scenarios are scientifically accurate. |
6.6 Humanities and Social Sciences |
Students in history, literature, and sociology must learn to interpret texts and analyze social phenomena. Generative AI can help by generating discussion questions and debate topics. For example, Perplexity can create a debate about a historical event from multiple perspectives. NotebookLM can be used to analyze primary sources. Instructors must ensure that AI does not flatten complex debates or introduce anachronisms. |
|
7. Comparing the Tools: Strengths and Weaknesses |
Perplexity and Google NotebookLM are not interchangeable. Each has strengths and weaknesses. |
Perplexity is strong at answering questions that require up-to-date information from the web. It provides citations, which helps students verify facts. However, it can sometimes produce answers that are too brief or too general. It can also be misled by low-quality sources. |
NotebookLM is strong at analyzing a specific set of documents. It is grounded in the user's own materials, which reduces the risk of hallucination. It can generate summaries, study guides, and even audio overviews. However, it is limited to the uploaded documents. If those documents are incomplete or biased, the AI's answers will be too. |
In practice, many instructors use both tools. Perplexity is used for exploration and fact-checking. NotebookLM is used for deep analysis of course materials. |
|
8. Survey-Based Evaluation: What the Data Shows |
Surveys conducted across multiple courses reveal several patterns. |
First, student engagement tends to increase when AI is used to generate realistic scenarios. Students report that they feel more motivated when they can see how abstract concepts apply to real-world problems. |
Second, acceptance levels vary. Some students embrace AI enthusiastically. Others are skeptical or anxious. Concerns include dependency, accuracy, and the fear that AI will replace human instructors. |
Third, the optimal pacing of AI integration is unclear. Some instructors introduce AI from the first day. Others wait until students have mastered the fundamentals. There is no consensus on which approach is better. |
Fourth, the balance between AI support and instructor-led guidance is delicate. Students appreciate AI for quick answers, but they also value the instructor's ability to provide context, nuance, and encouragement. Too much AI can make the course feel impersonal. Too little AI can make the course feel outdated. |
|
9. Ethical and Practical Considerations |
The use of generative AI in higher education raises several ethical and practical issues. |
Academic integrity is a major concern. If students can use AI to generate essays or solve problems, how can instructors assess their true understandingSome instructors respond by designing assessments that require original thinking and in-class work. Others embrace AI and ask students to critique its outputs. |
Privacy is another concern. NotebookLM requires uploading documents. If those documents contain sensitive information, such as student records or proprietary research, there is a risk of data leakage. Instructors must be careful about what they upload. |
Bias is a third concern. Generative AI models are trained on large datasets that may contain biases. If instructors are not careful, they may inadvertently reinforce stereotypes or present a narrow view of a subject. |
Access is a fourth concern. Not all students have equal access to AI tools. Some may lack reliable internet or adequate devices. Instructors must ensure that AI use does not disadvantage some students. |
|
10. Best Practices for Instructors |
Based on the experiences described in this chapter, several best practices emerge. |
First, start small. Introduce one AI tool for one specific task. See how students respond. Then expand gradually. |
Second, be transparent. Explain to students why you are using AI, what it can and cannot do, and how it will be assessed. |
Third, verify AI outputs. Do not assume that AI-generated content is accurate. Check facts, review scenarios, and remove anything that is misleading. |
Fourth, maintain balance. Use AI to supplement, not replace, instructor-led guidance. Keep the human element central. |
Fifth, encourage critical thinking. Ask students to critique AI outputs. This helps them develop the skills they need to use AI responsibly in their careers. |
Sixth, assess differently. Design assessments that value original thinking, problem-solving, and reflection. Avoid assessments that can be easily automated. |
|
11. Future Trajectories |
The use of generative AI in higher education is still in its early stages. Several trends are likely to shape the future. |
More personalized learning is likely. AI systems will become better at adapting to individual students' needs, providing tailored explanations, practice problems, and feedback. |
More integration with learning management systems is likely. AI tools will become embedded in platforms like Canvas and Moodle, making them easier for instructors and students to use. |
More emphasis on AI literacy is likely. Universities will increasingly offer courses on how AI works, how to use it ethically, and how to evaluate its outputs. This is already happening in computer science, but it will spread to other fields. |
More research on effectiveness is likely. As more data is collected, researchers will be able to identify which AI applications improve learning and which do not. This will help instructors make evidence-based decisions. |
More regulation is likely. Governments and accrediting bodies will develop guidelines for AI use in education. These guidelines will address issues such as privacy, bias, and academic integrity. |
|
12. Detailed Summary |
This chapter has explored the use of generative AI tools in higher education, with a focus on computer science, electrical engineering, and economics. It has also described applications in medicine, law, business, engineering, natural sciences, and humanities. |
The key findings are as follows. First, instructors are using tools like Perplexity and Google NotebookLM to customize learning experiences, design dynamic in-class activities, and generate realistic scenarios that bridge theory and practice. Second, survey-based evaluation reveals increased student engagement but varied acceptance levels. Third, concerns remain about the optimal pacing of AI integration and the balance between AI support and instructor-led guidance. |
The chapter has also compared the tools, discussed ethical and practical considerations, offered best practices, and outlined future trajectories. |
The central message is that generative AI is a powerful but imperfect tool. It can enhance higher education, but it cannot replace the human elements of teaching and learning. Instructors must use AI thoughtfully, critically, and ethically. They must also prepare students to use AI responsibly in their own careers. This is not just a technical challenge. It is a pedagogical and moral one. How we meet this challenge will shape the future of higher education for decades to come. |
In the end, the goal is not to automate education. The goal is to enrich it. Generative AI can help us do that, but only if we remain clear about our values and our purposes. The classroom is not a factory. It is a community of learners. AI can support that community, but it cannot replace it. That is the most important lesson of all. |
cs@easiersoft.com