Academically, Manus AI Can Be Discussed as an Example of Agentic Artificial Intelligence: Productivity, Responsibility, Digital Skills, Ethics, and the Changing Nature of Work

This article discusses Manus AI as an example of agentic artificial intelligence. Unlike traditional chatbots, which mainly answer prompts, agentic AI systems are designed to plan, act, use tools, and complete multi-step tasks with less direct human control. This shift is important for higher education because it changes how students learn, write, research, organize tasks, and prepare for the workplace. The article uses a conceptual academic method based on selected theories from sociology, education, and technology studies. It applies Pierre Bourdieu’s ideas of capital and habitus, world-systems theory, and institutional isomorphism to explain how agentic AI may affect students, universities, and labor markets. The analysis shows that agentic AI can increase productivity and support learning, but it also creates new questions about responsibility, academic integrity, digital inequality, dependence on platforms, and the value of human skills. The article argues that universities should not treat agentic AI only as a tool for cheating or automation. Instead, they should treat it as a new educational condition that requires updated teaching methods, assessment design, ethical rules, and digital skills training. The main finding is that agentic AI will not replace education, but it may change what it means to be educated. Students will need to learn how to ask good questions, check outputs, manage digital workflows, understand bias, protect privacy, and make responsible decisions when using AI agents.

Keywords: agentic artificial intelligence, Manus AI, higher education, digital skills, academic integrity, Bourdieu, world-systems theory, institutional isomorphism, future of work

1. Introduction

Artificial intelligence has moved from simple question-answer systems toward tools that can take action. In the first stage of public AI use, many students understood AI mainly as a chatbot. They asked a question, received an answer, edited the answer, and then decided how to use it. This model was already important because it changed writing, translation, summarizing, and study support. However, a new type of AI is now becoming more visible: agentic artificial intelligence.

Manus AI can be discussed academically as one example of this wider movement. The word “agentic” means that the system does not only respond to a single prompt. It can also plan steps, use tools, organize information, create outputs, and attempt to complete a task. In simple terms, a chatbot gives an answer, while an AI agent tries to do a job. This difference is not only technical. It is also educational, social, ethical, and economic.

For students, this development opens many questions. How should a student use an AI agent without losing their own learning? Who is responsible when an AI agent makes a mistake? What happens to academic integrity when a tool can produce reports, slides, websites, data analysis, or business plans? What kinds of digital skills will students need when AI systems can act across many platforms? Will students from richer countries and richer institutions gain more benefits than students from poorer contexts? Will universities copy each other’s AI policies because of pressure, ranking competition, and fear of being left behind?

These questions are not small. They show that agentic AI is not only a new educational technology. It is part of a broader change in knowledge production and work. Students are entering a world where tasks are increasingly shared between humans and digital systems. This does not mean that human intelligence becomes less important. It means that human intelligence must be used differently. The ability to judge, guide, question, verify, and ethically manage AI becomes more valuable.

This article aims to discuss Manus AI as a useful case for understanding agentic artificial intelligence in education. The article does not present Manus AI as the only or final model of AI agents. Instead, it uses Manus AI as an example of a wider shift from conversational AI to task-performing AI. The purpose is to build a simple but academically serious explanation of what this change may mean for students and universities.

The article is structured like a journal article. It begins with a background and theoretical framework. It then explains the method used in the discussion. After that, it analyzes the educational, social, and ethical meaning of agentic AI. The findings are then presented in a clear way. The conclusion argues that universities should move beyond fear-based reactions and develop responsible, practical, and human-centered approaches to agentic AI.

2. Background and Theoretical Framework

2.1 From Chatbots to AI Agents

Traditional chatbots are mainly conversational systems. They receive a prompt and produce a response. The user remains highly active in each step. The user asks, copies, edits, checks, and decides what to do next. This type of AI can support learning, but it usually depends on direct human prompting.

Agentic AI is different because it is designed around action. An AI agent may receive a goal and then divide that goal into smaller steps. It may collect information, compare options, create a document, build a plan, generate code, prepare a presentation, or organize a workflow. The user may still guide and check the process, but the system has more operational independence than a simple chatbot.

This difference matters in education. A student using a chatbot may ask for an explanation of a theory. A student using an AI agent may ask for a full research workflow: find themes, prepare an outline, create slides, draft a literature review, design a survey, clean a dataset, or prepare a project plan. This does not mean the work is automatically correct or ethical. It means the tool can now participate in more stages of academic work.

Agentic AI therefore changes the boundary between assistance and production. In older digital tools, students used software to type, calculate, design, or search. With AI agents, software may also suggest strategy, make decisions, and produce structured outputs. This creates a new educational problem: universities must decide what students are expected to do by themselves, what they may do with AI support, and what must be declared.

2.2 Bourdieu: Capital, Habitus, and Educational Advantage

Pierre Bourdieu’s theory is useful for understanding why agentic AI may benefit some students more than others. Bourdieu argued that education is not neutral. Students enter education with different forms of capital. Economic capital includes money and material resources. Cultural capital includes language skills, academic style, confidence, and knowledge of how institutions work. Social capital includes networks and useful relationships. Symbolic capital includes reputation and recognized status.

Agentic AI may become a new form of digital cultural capital. Students who know how to use AI agents well may complete tasks faster, write more clearly, prepare better presentations, and manage research more effectively. However, not all students will have the same ability to use these tools. Some students will have better devices, paid access, stronger English skills, better academic guidance, and more experience with digital platforms. Others may only have limited access or may not know how to judge the quality of AI outputs.

Bourdieu’s idea of habitus is also important. Habitus refers to the learned habits, expectations, and ways of acting that people develop through their social environment. A student whose habitus includes confidence with technology may use an AI agent as a normal academic tool. Another student may feel fear, confusion, or shame when using the same technology. Therefore, the impact of agentic AI is not only about access to a tool. It is also about the student’s background, confidence, language, and sense of belonging in academic spaces.

From a Bourdieusian view, agentic AI may reproduce inequality if universities do not provide clear training. Students with strong digital capital may gain more advantages, while students without such capital may fall further behind. But the opposite is also possible. If universities teach agentic AI skills fairly, these tools may help students who need support in writing, planning, translation, and organization. The result depends on institutional policy and teaching practice.

2.3 World-Systems Theory and Global Inequality

World-systems theory, associated with Immanuel Wallerstein, explains global society as a system divided into core, semi-peripheral, and peripheral regions. Core countries usually control advanced industries, capital, research systems, and high-value knowledge production. Peripheral countries often depend on imported technologies and external standards. Semi-peripheral countries stand between these positions.

Agentic AI can be studied through this theory because AI development is not equally distributed across the world. The strongest AI companies, cloud infrastructures, datasets, and computing resources are often concentrated in powerful economies. Students and universities in other regions may depend on platforms, languages, pricing models, and rules created elsewhere.

This creates a global education question. If agentic AI becomes central to study and work, then access to high-quality AI agents may become part of global academic inequality. Universities in wealthy countries may integrate advanced AI agents into teaching, research, administration, and career preparation. Universities with fewer resources may struggle with costs, training, regulation, and infrastructure. Students in different regions may therefore enter the future labor market with different levels of AI readiness.

Language is another issue. Many AI systems perform best in dominant global languages, especially English. This can benefit students who already have strong English skills. It can also create pressure on students from Arabic, Chinese, Spanish, French, Hindi, or other language backgrounds to work through English-speaking systems. Agentic AI may improve translation, but it may also strengthen the global power of certain languages and academic styles.

World-systems theory helps us avoid a narrow view of AI as a neutral tool. Agentic AI is part of a global system of technology, money, language, and institutional power. Its educational effects will depend on who controls the systems, who pays for them, who understands them, and whose knowledge is recognized as valuable.

2.4 Institutional Isomorphism and University Responses

Institutional isomorphism, developed by DiMaggio and Powell, explains why organizations often become similar to each other. They may copy each other because of rules, professional norms, or uncertainty. In higher education, universities often imitate policies, ranking strategies, quality assurance models, and digital systems.

Agentic AI may create strong isomorphic pressure. If leading universities adopt AI policies, AI literacy courses, AI research centers, or AI-based assessment tools, other universities may follow. Some may do so because of legal or accreditation pressure. Others may do so because they want to appear modern. Some may copy policies without fully understanding their local needs.

There are three common types of isomorphic pressure. Coercive pressure comes from laws, regulators, accreditation bodies, or government expectations. Normative pressure comes from professional standards, academic associations, and expert communities. Mimetic pressure happens when institutions copy others during uncertainty.

All three can appear in the case of agentic AI. Governments may demand AI ethics policies. Accreditation bodies may ask universities to show how they protect academic integrity. Professional communities may create guidelines for responsible AI use. Universities may copy well-known institutions because they do not want to seem outdated.

This theory helps explain why AI policy in universities may quickly become standardized. However, standardization is not always good. A copied policy may be too strict, too vague, or unsuitable for local students. A good AI policy should not only protect the institution. It should also help students learn how to use AI responsibly.

3. Method

This article uses a conceptual and interpretive method. It is not based on a survey, experiment, or interview study. Instead, it builds an academic discussion by connecting the example of Manus AI and agentic artificial intelligence with established theories in sociology, education, and technology studies.

The method has four parts.

First, the article defines agentic AI in simple educational terms. It focuses on the difference between systems that mainly respond and systems that can plan and act.

Second, the article treats Manus AI as an illustrative case. This means Manus AI is not studied as a full technical object, and the article does not make claims about every feature of the platform. Instead, Manus AI is used as a visible example of the broader movement toward AI agents that can support multi-step work.

Third, the article applies three theoretical lenses: Bourdieu’s theory of capital and habitus, world-systems theory, and institutional isomorphism. These theories are used because they explain inequality, global power, and institutional behavior.

Fourth, the article develops educational findings by analyzing how agentic AI may affect students, universities, academic integrity, digital skills, ethics, and work.

This method is suitable for an early-stage academic discussion because agentic AI is still developing. When a technology changes quickly, conceptual analysis can help universities ask better questions before large empirical evidence becomes available. However, this method also has limits. It cannot measure how many students use Manus AI or similar tools. It cannot prove learning outcomes. It cannot replace future empirical studies. Its purpose is to offer a structured academic interpretation.

4. Analysis

4.1 Productivity and the Student Experience

One of the clearest effects of agentic AI is productivity. Students often face many tasks at the same time. They must read, write, prepare slides, search for sources, manage deadlines, communicate with teachers, and sometimes work part-time. An AI agent can help organize some of these tasks. It may create a study plan, summarize materials, prepare a first draft, organize notes, or produce a presentation structure.

This can be useful, especially for students who struggle with planning. Many students do not fail because they lack intelligence. They fail because they cannot manage time, break tasks into steps, or understand academic expectations. Agentic AI may help by making the process more visible. It can show a possible sequence: define the topic, collect sources, build an outline, write sections, check arguments, and revise.

However, productivity is not the same as learning. A student may finish a task faster but understand less. If an AI agent does too much, the student may become a manager of outputs rather than a learner. This is a serious educational concern. Learning often happens through difficulty. Students learn by searching, making mistakes, comparing ideas, and revising their own work. If the agent removes all difficulty, it may also remove part of the learning process.

The main question is therefore not whether agentic AI increases productivity. It likely can. The better question is: productivity for what purpose? If the goal is to avoid thinking, the tool harms education. If the goal is to reduce unnecessary workload and help students focus on judgment, reflection, and deeper understanding, the tool can support education.

Universities should help students distinguish between useful support and harmful dependence. For example, using an AI agent to organize a reading schedule may be acceptable. Asking it to write an entire assignment without understanding the content is not acceptable. Asking it to generate possible counterarguments may be educational. Submitting its output as personal work without declaration is dishonest.

4.2 Responsibility and Human Oversight

Agentic AI creates a responsibility problem. When a chatbot gives a wrong answer, the user can often see that it is only a response. When an AI agent completes a full task, the output may appear more finished and professional. This may create false confidence. Students may trust the result because it looks polished.

But a polished output can still contain errors. It may include weak arguments, invented facts, biased assumptions, incorrect calculations, or unsuitable sources. If a student submits such work, the responsibility cannot be placed only on the AI system. In education, the student remains responsible for submitted work.

This means that AI literacy must include verification. Students need to learn how to check facts, compare sources, test arguments, review calculations, and identify unsupported claims. They also need to understand that AI systems do not have moral responsibility in the human sense. The tool can produce text or actions, but the student must decide whether the result is accurate, ethical, and appropriate.

The issue becomes more complex when AI agents act across tools. If an agent sends an email, creates a file, changes data, or makes a decision based on instructions, mistakes can have real effects. In academic life, this may include sending wrong information, misusing data, or producing misleading research materials. In professional life, the risks may be even higher.

Therefore, human oversight should be a core rule. Students should not be trained to obey AI agents. They should be trained to supervise them. Supervision requires knowledge. A person cannot check an AI-generated statistical analysis without basic statistical understanding. A person cannot judge a legal or ethical argument without some knowledge of the field. This is why agentic AI does not remove the need for education. It increases the need for strong foundational learning.

4.3 Digital Skills as a New Academic Requirement

Digital skills used to mean basic computer use: typing, searching, email, spreadsheets, and presentation software. In the age of agentic AI, digital skills are broader. Students need to understand how to design prompts, set goals, break down tasks, evaluate outputs, protect data, and document their use of AI.

A student who uses agentic AI well must know how to communicate clearly with the system. Poor instructions produce poor results. Clear instructions require the student to understand the task. This means that prompt writing is not only a technical skill. It is also a thinking skill. The student must define purpose, audience, format, evidence, tone, and limits.

Students also need workflow skills. Agentic AI may support complex tasks, but students must know how to organize the process. They must decide when to use AI, when to work alone, when to ask a teacher, and when to stop. They must also keep records. In academic settings, students may need to declare how AI was used. This requires a habit of documentation.

Digital ethics is another skill. Students must know what kind of information should not be uploaded into AI systems. Personal data, confidential documents, unpublished research, and sensitive institutional materials require protection. Students must understand that convenience does not remove privacy obligations.

Digital skills also include the ability to work with uncertainty. AI outputs may be useful but incomplete. Students must learn to ask: What is missing? What is assumed? What evidence is needed? What could be wrong? This type of critical questioning is a key academic skill.

4.4 Academic Integrity and Assessment

Agentic AI challenges traditional assessment. Many universities still use essays, reports, and take-home assignments as evidence of student learning. But if AI agents can produce these outputs, universities must rethink how they assess knowledge.

This does not mean essays are useless. Writing remains important because it teaches structure, argument, and reflection. But universities may need to change how writing is taught and assessed. Teachers may ask students to submit drafts, reflection notes, oral defenses, annotated sources, or AI-use declarations. Assessment may focus more on process, reasoning, and personal understanding.

Academic integrity policies should be clear but not extreme. A total ban on AI may be unrealistic and may punish honest students while dishonest students continue using tools secretly. On the other hand, unlimited use of AI may weaken learning and fairness. A balanced policy is better. It should explain which uses are allowed, which uses must be declared, and which uses are forbidden.

For example, AI may be allowed for brainstorming, grammar support, translation support, or planning. It may be allowed for generating practice questions. It may be allowed for coding support if declared. But submitting AI-generated work as fully personal work should not be accepted. Using AI to invent references, falsify data, or avoid required learning should be treated as misconduct.

Agentic AI also raises the question of authorship. If an AI agent creates a major part of an assignment, can the student be considered the author? In academic terms, authorship requires responsibility. The author must understand, defend, and take responsibility for the work. Therefore, even if AI is used, the student must be able to explain the ideas, methods, and conclusions.

Oral assessment may become more important. A short oral discussion can show whether the student understands the submitted work. Project-based learning may also become more useful because it allows teachers to observe the process. Universities may also design assignments that require local context, personal reflection, field observation, or direct application, making simple automation less useful.

4.5 Ethics: Bias, Privacy, and Dependence

Ethics is central to the use of agentic AI. The first issue is bias. AI systems learn from data and patterns. These patterns may reflect social inequalities, cultural assumptions, or dominant worldviews. If students use AI outputs without criticism, they may reproduce these biases.

The second issue is privacy. AI agents may need access to files, browsers, emails, or documents to complete tasks. This creates risks. Students may upload private information without understanding where it goes or how it may be processed. Universities must teach students to protect sensitive information.

The third issue is dependence. If students rely too much on AI agents, they may lose confidence in their own thinking. They may feel unable to write, plan, or solve problems without digital support. This is not only an academic issue. It is also a psychological and professional issue. Education should build independence, not only efficiency.

The fourth issue is transparency. Students should know when and how AI has been used. Teachers should also be transparent when they use AI in teaching or grading. Hidden AI use can reduce trust. Clear rules can protect both students and institutions.

The fifth issue is fairness. If some students can pay for advanced AI agents and others cannot, assessment may become unequal. Universities should consider whether they need to provide institutional access or design tasks that do not reward paid tools unfairly.

4.6 Bourdieu and the Unequal Value of AI Skills

Using Bourdieu, agentic AI can be understood as a new field of educational advantage. Students who know how to use AI agents may gain a form of digital cultural capital. They can produce better-looking work, communicate more professionally, and complete tasks faster.

However, this advantage is not only about the tool. It is about knowing how to use the tool in a way that matches academic expectations. A student from an educated family may already know how to structure an essay, write formal English, and communicate with professors. AI can strengthen these existing advantages. Another student may use the same tool but still produce weak work because they do not understand academic norms.

This shows why universities should not assume that AI access creates equality. Access is only the first step. Students need guided practice. They need examples of good and bad AI use. They need feedback. They need to understand academic style, evidence, and argument.

Bourdieu also helps explain symbolic capital. Students who use AI well may appear more professional and capable. Their work may look polished. But polished appearance can hide weak understanding. Universities must therefore assess depth, not only presentation quality.

4.7 World-Systems Theory and the Global AI Divide

World-systems theory shows that agentic AI may deepen global educational inequality if access and control remain concentrated. AI platforms are often produced by powerful companies and advanced economies. Their design may reflect the priorities of these centers of power. Students and universities outside these centers may become consumers of systems they do not control.

This matters for curriculum. If universities simply import AI tools without local adaptation, they may also import assumptions about language, work, knowledge, and success. Local knowledge may become less visible. Local academic traditions may be pushed aside by global templates.

At the same time, agentic AI may also create opportunities for semi-peripheral and peripheral institutions. Students in smaller universities may gain access to advanced support that was previously unavailable. They may use AI agents for translation, research organization, coding, design, and international communication. This can help reduce some gaps.

The outcome depends on policy. If AI is expensive, closed, and controlled by a few actors, inequality may grow. If AI literacy is taught widely and tools become accessible, more students may benefit. Universities in different regions should develop their own AI strategies instead of only copying models from elite institutions.

4.8 Institutional Isomorphism and AI Policy

Universities are already under pressure to respond to AI. Some create AI policies quickly because they fear cheating. Others create AI centers because they want to appear innovative. Some adopt detection tools. Others promote AI literacy. Over time, many universities may start to look similar in their AI responses.

Institutional isomorphism explains this pattern. During uncertainty, organizations copy each other. This can be helpful when good practices spread. But it can also lead to shallow policies. A university may publish an AI policy without training teachers. It may buy software without changing assessment. It may use strong ethical language without practical support for students.

A better approach is reflective adoption. Universities should ask: What do our students need? What are our risks? What are our values? What resources do we have? How can we protect integrity while supporting learning?

Agentic AI policies should be practical. They should include clear examples. They should explain allowed, limited, and forbidden uses. They should guide teachers on assessment design. They should protect privacy. They should support students who lack digital skills.

4.9 The Changing Nature of Work

Agentic AI is important because students are not only preparing for exams. They are preparing for work. Many jobs include repeated tasks, information processing, writing, planning, customer communication, reporting, and digital coordination. AI agents may automate parts of these tasks.

This does not mean all jobs will disappear. Work usually changes before it disappears. Some tasks are automated, while new tasks appear. Workers may need to supervise AI systems, check outputs, manage exceptions, communicate with clients, and make ethical decisions. The human role may move from doing every step to managing complex human-AI workflows.

Students therefore need more than technical training. They need judgment, communication, ethical awareness, creativity, and adaptability. They must know how to work with AI without becoming passive. They must understand both the power and the limits of automation.

In this sense, agentic AI may change employability. Employers may expect graduates to know how to use AI agents responsibly. A student who refuses to learn these tools may be disadvantaged. But a student who uses them without understanding may also be risky. The best graduate will be one who can combine human judgment with digital capability.

5. Findings

This conceptual analysis leads to seven main findings.

Finding 1: Agentic AI changes the meaning of digital assistance

Manus AI and similar systems show that AI assistance is moving from answering questions to completing tasks. This changes the student’s relationship with technology. The student is no longer only asking for information. The student may be delegating parts of academic or professional work.

Finding 2: Productivity gains are real but educationally incomplete

Agentic AI can help students save time, organize work, and produce outputs. However, faster production does not automatically mean deeper learning. Universities must help students use AI for learning, not only for task completion.

Finding 3: Responsibility remains human

Even when an AI agent produces a polished output, the student remains responsible for accuracy, honesty, and ethical use. Human oversight is essential. Students must be able to explain and defend any work they submit.

Finding 4: AI literacy is becoming part of academic literacy

Students now need skills in prompting, verification, workflow design, privacy protection, and ethical judgment. These skills should be taught as part of modern higher education.

Finding 5: Agentic AI may reproduce inequality

Using Bourdieu’s theory, AI skills can become a new form of cultural capital. Students with better access, stronger language skills, and more digital confidence may benefit more. Without training, AI may widen gaps between students.

Finding 6: Agentic AI reflects global power structures

World-systems theory shows that AI tools are part of global inequalities in technology, language, infrastructure, and knowledge production. Universities outside the global core should develop local AI strategies rather than only importing external models.

Finding 7: Universities may copy AI policies without deep change

Institutional isomorphism explains why universities may quickly adopt similar AI policies. But policy copying is not enough. Real change requires teacher training, assessment redesign, student support, and ethical governance.

6. Conclusion

Manus AI can be discussed academically as an example of agentic artificial intelligence because it represents a shift from AI that mainly responds to AI that can plan and act. This shift matters for students because it changes study habits, academic responsibility, digital skills, ethics, and preparation for work.

The main educational challenge is not simply whether students should use AI agents. They already will. The real challenge is how to guide their use in a responsible and meaningful way. Universities should not respond only with fear. They should also not accept agentic AI without rules. A balanced approach is needed.

This article argued that Bourdieu helps explain how agentic AI may become a new form of educational advantage. World-systems theory shows that AI is connected to global inequalities in technology and knowledge. Institutional isomorphism explains why universities may copy AI policies during uncertainty. Together, these theories show that agentic AI is not just a technical tool. It is a social and educational force.

For students, the future will require more than the ability to use software. They will need to manage AI agents, check outputs, understand bias, protect data, and make responsible decisions. They will need to prove not only that they can produce work, but that they understand it.

For universities, the task is clear. They should teach AI literacy, redesign assessment, protect academic integrity, support equal access, and build ethical rules that are practical. Agentic AI should not replace education. It should push education to become more reflective, more responsible, and more connected to the changing world of work.

The most important point is simple: as AI becomes more agentic, human judgment becomes more important, not less. The value of education will depend on helping students become thoughtful supervisors of technology, not passive users of it.

Hashtags

#AgenticArtificialIntelligence #ManusAI #HigherEducation #DigitalSkills #AcademicIntegrity #EthicalAI #FutureOfWork #AIInEducation #StudentLearning #ResponsibleTechnology

References

• Bourdieu, P. (1977). Outline of a Theory of Practice. Cambridge University Press.

• Bourdieu, P. (1984). Distinction: A Social Critique of the Judgement of Taste. Harvard University Press.

• Bourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education. Greenwood Press.

• Bourdieu, P. (1991). Language and Symbolic Power. Harvard University Press.

• Brynjolfsson, E., & McAfee, A. (2014). The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. W. W. Norton.

• Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press.

• DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160.

• Floridi, L. (2014). The Fourth Revolution: How the Infosphere is Reshaping Human Reality. Oxford University Press.

• Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., Luetge, C., Madelin, R., Pagallo, U., Rossi, F., Schafer, B., Valcke, P., & Vayena, E. (2018). AI4People: An ethical framework for a good AI society. Minds and Machines, 28, 689–707.

• Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2), 340–363.

• Nwana, H. S. (1996). Software agents: An overview. The Knowledge Engineering Review, 11(3), 205–244.

• Russell, S., & Norvig, P. (2021). Artificial Intelligence: A Modern Approach (4th ed.). Pearson.

• Selwyn, N. (2016). Education and Technology: Key Issues and Debates. Bloomsbury Academic.

• Selwyn, N. (2019). Should Robots Replace Teachers? AI and the Future of Education. Polity Press.

• Suchman, L. (2007). Human-Machine Reconfigurations: Plans and Situated Actions. Cambridge University Press.

• Wallerstein, I. (1974). The Modern World-System I: Capitalist Agriculture and the Origins of the European World-Economy in the Sixteenth Century. Academic Press.

• Wallerstein, I. (2004). World-Systems Analysis: An Introduction. Duke University Press.

• Wooldridge, M. (2009). An Introduction to MultiAgent Systems (2nd ed.). Wiley.

• Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs.