Deng Guomin 1, Li Mei 2

(1. School of Educational Science, Guiyang University, Guiyang, Guizhou; 2. School of Education, Beijing Open University, Beijing)

The article aims to reveal what new ethical problems will be faced in education in the intelligent era, and explore the ethical principles of educational artificial intelligence based on artificial intelligence ethics and educational ethics. Using literature research and logical reasoning methods, we systematically sorted out the characteristics and advantages of the application of educational artificial intelligence and the new ethical problems that may arouse: (1) At the role level, stakeholders such as system creators, teachers, students and monitors will face more complex ethical dilemmas; (2) At the technical level, it involves issues such as ethical accountability for automated decision-making, algorithmic ethics and data ethics; (3) At the social level, it may transform the employment market, change interpersonal relationships and cause new social injustice. Therefore, it is necessary to have a full understanding of the complexity of the ethical issues of educational artificial intelligence, and to follow general ethical principles of artificial intelligence and educational ethical principles. Among them, the most core content is the intersection between the two, including: (1) welfare; (2) right and wrong, good and evil; (3) fairness and justice; (4) human rights and dignity; (5) freedom and autonomy; (6) responsibility and accountability.

Keywords

Educational Artificial Intelligence; Artificial Intelligence Ethics; Educational Ethics; Data Ethics; Algorithmic Ethics; Ethical Principles

1. Introduction

At present, the development of new technologies such as artificial intelligence, big data and the Internet of Things is promoting human society from the information age to the intelligent era. Education in the information age still has obvious industrial characteristics: with the help of modern information technology, the efficiency of creating, transmitting and sharing of educational resources will be improved, the limitations of time and space will be broken, and the batch and efficiency of talent training will be maximized. However, the teacher resources in the information age are still short of, and teachers are deeply trapped in a large number of trivial daily work, making it difficult to achieve true teaching according to their aptitude.

Educational artificial intelligence is an artificial intelligence technology and system applied in the field of education. With its highly intelligent, automated and high-precision data analysis and processing capabilities and active learning capabilities, it can undertake some of the tasks that only teachers can be competent in the past, thereby effectively solving a series of problems faced by education in the information age: (1) Using automatic question-and-answer technology based on natural language processing and deep learning, it can automatically answer questions raised by students; (2) Automatically evaluate students' learning and homework, and provide feedback and support that conforms to their cognitive status; (3) Using emotion detection technology to automatically identify students' body movements and facial expressions, identify students' emotional needs during the learning process, and provide personalized emotional support.

;(4) Simulate different roles such as teachers, managers, learning partners, collaborators and competitors to promote communication and cooperation

; (5) Build an adaptive learning environment, analyze and diagnose learners' learning styles and learning needs, and provide them with adaptive learning support services; (6) Use decision-making management technology to assist teachers' teaching decisions and learners' learning decisions

. Therefore, educational artificial intelligence can support personalized teaching more effectively, and even its ability to support services in some learning has surpassed human teachers.

However, the strong data integration and analysis capabilities of educational AI will also raise a series of new ethical issues, such as: it may pose potential risks to users' private data and privacy protection; blur the boundaries between roles such as teachers, students and AI agents. Therefore, the design, development and teaching practice application of educational artificial intelligence all require a full understanding of the complexity of its ethical issues and carefully consider what ethical principles should be followed.

2. Ethical problems faced by educational artificial intelligence

(I) Ethical issues of stakeholder

To explore the ethical issues of educational artificial intelligence, we need to determine which stakeholders it involves. From the perspective of interpretability or transparency of artificial intelligence, Tomset () and others define six different roles in the artificial intelligence ecosystem, namely system creator, system operator, decision executor, decision-making subject, data subject and system monitor.

. Educational artificial intelligence is an artificial intelligence technology and system applied in the field of education, and its stakeholders should also include these roles. First, the creators include software companies, employees, and relevant professionals and researchers who work with them. Secondly, both teachers and students assume the role of operators, executors, decision-makers and data subjects, because in the teaching process, teachers play a leading role and students have a subjective position. They will directly interact with the educational artificial intelligence to provide the system with input and receive information output from the system. Teachers and students will make decisions based on the information output from educational artificial intelligence, and will be affected by the other party's decisions. Their personal information may be used to train educational AI, enrich and perfect their intelligent tutor agents, learner models and pedagogy models. Finally, monitors are agencies or personnel that review educational artificial intelligence, who review the effectiveness of educational artificial intelligence and the potential risks that may arise. Therefore, in the process of building educational artificial intelligence, it is necessary to consider the ethical issues of stakeholders such as system creators, teachers, students and monitors.

1. System creator ethics issues

The creators of educational artificial intelligence include system designers, software architects, programmers and other design developers. In the process of designing and developing educational artificial intelligence, they need to take into account the interests of teachers and students while taking into account the benefits of their teaching applications. They need to support the subjective initiative of teachers and students while ensuring the robustness, security, interpretability, non-discrimination and fairness of educational artificial intelligence, attach importance to the physical and mental health of teachers and students and the comprehensive development of students' comprehensive quality, and need to be able to support accountability and accountability. To meet these requirements, the creators of educational artificial intelligence should be clear about what ethical responsibilities they bear. First of all, designers need to understand the differences in abilities, needs, and social culture between different users, and truly design products based on good motives to avoid any form of harm and discrimination. Software architects need to build a safe, reliable and transparent system architecture system to better support accountability tracking while protecting the security of user personal data and ensuring the reliable operation of the system. For programmers, it is necessary to design unbiased algorithms so that they can treat every user fairly and impartially.

2. Teacher ethics issues

Artificial intelligence technology will automate many professions. A large part of teachers' daily work, such as teaching evaluation, teaching diagnosis, teaching feedback and teaching decisions, will be shared by educational artificial intelligence, and the role of teachers will continue to deepen and eventually change. On the one hand, teachers will be liberated from some repetitive and trivial daily affairs and will have more time and energy to devote themselves to more creative, humane and critical teaching work, such as creating and deploying a more complex and intelligent learning environment, utilizing and expanding their own professional knowledge, and providing learners with more targeted support at the right time. On the other hand, the importance of technology will have a direct impact on the teacher's work and role. For example, it may cover up the importance of the teacher's personal abilities, knowledge and personality, and weaken the teacher's subjective initiative.

. Therefore, in an intelligent teaching environment, teachers will face a series of new ethical dilemmas: (1) How to define and cope with their new role status and their relationship with intelligent tutors; (2) How to protect their privacy and personal security in the process of using educational artificial intelligence to collect and use students' learning behavior data to provide teaching support services; (3) In the new teaching practice environment, what kind of ethical knowledge should teachers have and what ethical standards should follow to make ethical decisions. Therefore, educational artificial intelligence will put forward more new requirements for teachers' abilities, qualities and knowledge structure. They not only need to have certain abilities and knowledge about the design, development and application of educational artificial intelligence, but also need to have certain ethical knowledge so that they can integrate educational artificial intelligence into educational practice in a way that conforms to ethical principles.

3. Student ethics issues

The era of intelligence has put forward more new requirements for students' learning, ability and literacy, and has also made the ethical problems faced by students more complicated. First, artificial intelligence may change the future employment market, and employment-oriented education is likely to be outdated. Therefore, learners need to rethink the content and goals they learn.

. Second, the human nervous system is plastic. Educational artificial intelligence can provide learners with personalized learning support services, but it may also deprive them of many opportunities for thinking training and learning experiences, changing their brain structure and cognitive abilities, especially during the critical period of learners' brain development, overuse of cognitive technology may have irreversible serious consequences, which is very unfavorable for them to develop more advanced cognitive skills and innovative abilities.

. Third, the development of educational artificial intelligence has made ethical issues such as intellectual property rights and integrity more subtle and complex. If students use educational artificial intelligence to automatically generate papers, homework or works, and the issue of the ownership of intellectual property rights involved in it still lacks clear ethical norms.

10

. Fourth, will the use of educational artificial intelligence bring repetitive strain, vision problems and obesity problems to learners, will it affect their senses such as vision, touch and hearing, and will it deprive them of the opportunity to experience a simple and happy life.

11

. Fifth, educational artificial intelligence may deprive a large number of learners of the opportunity to communicate with teachers and peers, and will this affect their ability to participate in meaningful human social activities. Sixth, the cognitive ability, computing ability and information processing ability of educational artificial intelligence far exceed that of human learners in some aspects. Will it make them frustrated, reduce their learning motivation, and make them no longer willing to actively solve problems and think rationally.

4. Monitor ethical issues

In the educational artificial intelligence scenario, monitors include institutions or relevant personnel, data specialists, etc. that formulate educational artificial intelligence standards. They are mainly responsible for testing, investigating and reviewing the consistency, security and fairness of the system. They are mainly responsible for testing, investigating and reviewing the consistency, security and fairness of the system.

. Security testing is generally performed before the deployment of educational AI, and review requires checking the system's past output, so a backup system is needed to store educational AI data, decisions and explanations. It is through these data and interpretations that monitors interact with educational artificial intelligence, analyze and review the consistency, safety and fairness of educational artificial intelligence's decisions on teaching support services, and propose corresponding rectification measures. Therefore, to ensure the interpretability of educational artificial intelligence, it is necessary to establish corresponding storage retention mechanisms and interpretation plans for the decisions, judgments and learning support services they make for review by monitors, and to ensure that the design, development and application of educational artificial intelligence is in line with the ethical norms of education and teaching through third-party institutions. After the review, monitors also need to provide creators with feedback on how to improve educational AI, such as retraining models, improving model equity, and security strategies to protect students’ privacy data.

(II) Technical ethical issues

1. Issues of ethical accountability for automated decision-making

As educational artificial intelligence continues to increase their ability to control and make decisions, they also need to be aware of the ethical implications of their own behavior rather than relying on rules preset by designers. Therefore, it is necessary to introduce ethical principles and rules to educational artificial intelligence, build an intelligent moral agent, so that it has certain ethical decision-making capabilities, and be able to judge its own decisions, behaviors and consequences, and coordinate the interests of all parties, but this will also cause some new ethical problems

12

. First of all, rules and algorithms are customized by designers and programmers, rather than independently constructed by educational artificial intelligence. Authority, power and influence will increasingly be reflected in the development and implementation of algorithms. In the case of conflicts of interest between all parties, whose interests should educational artificial intelligence be prioritized, designers, programmers, teachers or students? Secondly, the application of educational artificial intelligence may lead to learning failures, among which the potential responsible parties include design developers, teachers, students and intelligent agents. Who should be held accountable when this happens? What responsibilities should software, data, teachers and students bear? There is currently a lack of corresponding accountability mechanisms. Finally, the automation and complexity of educational artificial intelligence itself means it is difficult for us to determine how the decision-making mechanism of the system operates, and it is difficult to trace and explain it.

2. Algorithm ethics issues

At present, artificial intelligence mainly learns and trains through machine learning and deep learning algorithms, and makes decisions and judgments based on historical data recorded in the past. If educational artificial intelligence is used to evaluate learners and their learning, it can only be carried out based on past standards and methods, such as success measures in a specific historical and cultural context. This will inevitably lead to the learning algorithms that produce historical or cultural biases and biases, and are limited to processing based on averaged data of historical cases, and face great challenges in dealing with creative and innovative learning activities.

. From a basic ethical point of view, the expression of initiative of teachers and students requires not only repeating past historical choices, but also emphasizing people's creativity and innovation, and cannot just stay in the past. Educational artificial intelligence may provide more personalized and diverse learning paths and learning resources options, but the current practice is generally to identify users as different personal portraits based on their past choices and decisions, and then filter out learning content and paths that do not match their personal portraits. This approach may limit learners’ need for diversified learning content and ignore the flexibility, diversity and uncertainty of human learning and development.

3. Data ethics issues

The normal operation of educational artificial intelligence requires the support of a large amount of data, so it is necessary to record a large amount of data on learners' abilities, emotions, strategies and misunderstandings, involving a large number of new data ethics issues. First of all, whether there will be an unexpected data leakage in the analysis and processing of data will be involved, and to what extent should teachers and students control their own data and protect themselves from harm and discrimination. Secondly, whoever owns this data, what privacy issues are involved, and who should be responsible for it if personal information is leaked. Finally, for these data, how should be analyzed, interpreted and shared, and how to correct biases (consciously and unconsciously) that may negatively affect individual students’ legal rights, especially in consideration of race, social status, and income inequality. Therefore, the ethical principles of educational artificial intelligence cannot be simplified to questions about data or calculation methods, and data ethics and educational ethics need to be considered. The ethical issue of educational artificial intelligence is actually a comprehensive application of data ethics, computing ethics and educational ethics.

13

(III) Social ethical issues

1. Change the direction of employment market and talent training

One of the key roles of the current education system is to cultivate learners' ability to participate in socio-economic life. The history of the education system is closely related to the development of industrial society, and wage labor remains the core organizational principle of today's society and its daily life. However, artificial intelligence will likely replace many human jobs. Will this lead to an increase in the unemployment rate? Looking at the industrial revolutions, similar concerns will arise every time, but in the end, the demand for talents will become increasingly greater, because with the growth of population and the development of society, new demands will be put forward for products and services.

. Some simple and repetitive labor will be undertaken by artificial intelligence, while humans will engage in some creative, challenging and irreplaceable tasks, such as the design and development of artificial intelligence systems, research on philosophical issues of artificial intelligence and social issues. They will have more time and energy to devote themselves to the areas they are interested in and truly realize the liberation of human nature. Therefore, the future employment market structure will face a fundamental change, and society will also put forward more new requirements for the ability and literacy of talents, which will have a revolutionary impact on the quality standards and specifications of talent training.

2. Change interpersonal relationships during teaching

The technical process of educational artificial intelligence may ignore the correct relationship between teachers and students and between students (“I-you” rather than “I-he”), weakening our understanding of education, and then simply understanding it as learning skills rather than human cultivation, so as to easily ignore the importance of teachers and teaching in the educational process.

14

. Educational artificial intelligence will have an important impact on teacher-student relationships, which may weaken the level of connection between teachers and students and their ability to establish connections, while also potentially weaken the teaching and learning process and weaken the status of teachers as professionals. The quality of teacher-student relationships has an important impact on learners' personal self-esteem, learning motivation, and confidence in challenging new problems. Teachers’ friendly attitude and timely feedback to students help them gain self-confidence and promote the development of their healthy emotions. Moreover, we cannot guarantee that all educational AI users are based on a good intention. The abuse and bad use of educational AI, such as obtaining other people's privacy data, imitating other people's handwriting, voice and conversational styles to obtain illegal benefits, and taking the knowledge, reports and art works generated by educational AI for themselves, may reduce the sense of trust between people in the teaching process.

3. Cause new social injustice

The original intention of educational artificial intelligence is to enhance the welfare of all stakeholders in the education system and benefit more teachers and students. New social injustice caused by the digital divide need to be avoided as much as possible. The digital divide in the field of educational artificial intelligence is not simply caused by the huge difference in ownership and access rights of technological equipment caused by differences in socio-economic status, but is composed of a series of complex influencing factors, such as information literacy, artificial intelligence literacy, attitude towards educational artificial intelligence, application ability and intention to use, etc. They will affect the purpose, methods and strategies of teachers and students to apply educational artificial intelligence, and thus affect the final teaching and learning effect, bringing new educational injustice.

. Historical experience also shows that the application of new technologies often does not narrow the inequality of education, but may expand the inequality of education, which involves issues such as the ability, effect and efficiency of technology application.

3. Ethical principles of educational artificial intelligence

The automation, autonomy and initiative of artificial intelligence itself make it play the role of moral actors and recipients at the same time, and has certain ethical responsibilities and moral decision-making capabilities. The application of educational artificial intelligence may put various stakeholders in many new ethical dilemmas and bring about a series of new technological and social ethical dilemmas. Therefore, the application of educational artificial intelligence is not only a technical issue, but also involves issues at the relationship between technology and humanity, science and technology philosophy and sociology. If you do not have a deep understanding of the complexity of its ethical dilemma, it may have immeasurable negative impacts. On the one hand, educational artificial intelligence is the specific application of artificial intelligence technology and systems in the field of education, and needs to follow general artificial intelligence ethical principles; on the other hand, it is a specific educational and teaching practice activity, which needs to follow educational ethical principles.

According to the International Association of Electrical and Electronic Engineers (IEEE) 2019 "Ethical Design: Vision to Maximize Human Well-being with Artificial Intelligence and Autonomous Systems (First Edition)"

15

, Analysis of the "Ethical Guidelines for Artificial Intelligence that Reliable" issued by the European Commission in 2019 and other AI ethical guidelines documents

16

, it can be found that although there are differences in cultural traditions among countries, it shows certain convergence in general artificial intelligence ethics, mainly including interpretability, fairness and justice, responsibility, privacy and data governance, kindness, non-malice, freedom and autonomy, trust, dignity, robustness, welfare, effectiveness, prudence, competence and human rights. The principles of education ethics mainly include: (1) Whether it can bring welfare to all stakeholders of educational activities; (2) Whether it can distinguish right from wrong, good from evil; (3) Whether it respects educational fairness and educational justice; (4) Whether it protects basic human rights such as education rights, freedom and privacy rights; (5) Whether it respects people and maintains human moral dignity; (6) Whether it fulfills the moral responsibilities and obligations of education

17

. It can be seen that there are overlaps between artificial intelligence ethics and educational ethics in terms of welfare, right and wrong, good and evil, fairness and justice, human rights and dignity, freedom and autonomy and responsibility, as shown in Figure 1.

Figure 1 The relationship between artificial intelligence ethics and educational ethics

Therefore, the ethical principles of educational artificial intelligence should be the intersection between the two, representing the specific ethical principles that should be followed in the technological development and practical application of artificial intelligence systems in the educational situation.

(I) Welfare

Welfare mainly refers to the design and application of artificial intelligence that should improve human quality of life and external environment, including various personal, social and environmental factors on which humans rely on to survive. It needs to evaluate whether artificial intelligence will make human society and environment better, whether it can promote social progress, promote sustainable development of society and environment, own and benefit people everywhere

15

. From the perspective of the education field, welfare mainly refers to the design, development and application of educational artificial intelligence should be able to bring benefits to all stakeholders of educational activities and maximize the educational interests of all personnel involved in educational and teaching activities. For example: Can it improve the existing teaching environment and better support learners' learning, especially those with learning disabilities; can it provide more and better learning opportunities, better teach students according to their aptitude, so that all learners can develop more fully; can it liberate teachers from some low-level repetitive work, so that they can better invest in more valuable, meaningful and creative educational activities, and make their personal value more fully realize; can it improve the relationship between teachers and students so that teachers and students can devote themselves to teaching and learning activities in a happier and healthier way, and achieve comprehensive and harmonious development of individuals while better protecting their physical and mental health.

(II) Right and wrong, good and evil

The non-malice principle of artificial intelligence mainly refers to the fact that its design and application should be based on a desire for benevolence rather than maliciousness, that is, it should never cause foreseeable or unconscious harm, prejudice and injustice, and that specific risks or potential harms need to be avoided.

18

. Therefore, the design, development and deployment of artificial intelligence products need to be based on a good desire, and also requires a set of intervention mechanisms to prevent risks and potential hazards. Educational ethics emphasizes that educational behavior needs to be based on good motives, follow moral norms with universal value and lofty moral significance, and fulfill social and individual moral obligations. Therefore, the design and development of educational artificial intelligence should be based on a kind motivation, which is to truly achieve better development of education, to enable teachers to better carry out teaching work and fulfill their social responsibilities in educating people, and to better learn and promote their comprehensive and harmonious development, rather than for some purpose that is utilitarian or even personal desires.

(III) Fairness and justice

The principle of fairness and justice requires that AI is fair and just for all, and that existing social biases should be avoided and new ones should be introduced. Therefore, AI decision algorithms and training data should not follow the designer's inherent biases and discrimination, but should ensure that these data are appropriately representative of different socio-cultures, giving all the opportunity to access equally by assessing them

19

. Educational and teaching activities need to maintain the fairness and justice of education. The right to education is the basic right of every citizen in society, regardless of the level of his social status and the amount of wealth. Moreover, education itself is also a way to achieve social equity, which provides a good channel for the flow of classes.

20

. Before deploying an educational artificial intelligence system, you need to carefully consider whether it will bring more educational opportunities, whether it will promote educational equity, or hinder educational equity. First of all, whether there is bias and discrimination in the decision-making algorithms and data that are based on education, and whether every student can be treated fairly and impartially. Secondly, will the difference in the level of teachers' ability to use educational artificial intelligence to teach lead to differences in teaching effects, thereby causing new educational injustice. Again, the difference in the information literacy and artificial intelligence literacy levels of learners will bring about differences in learning effects on whether using educational artificial intelligence for learning will lead to a difference in learning effects, thereby widening the gap between students' academic performance. Finally, will the difference in family social status and economic strength cause unequal opportunities for accessing education artificial intelligence, thus leading to new educational inequalities?

(IV) Human Rights and Dignity

Artificial intelligence should prioritize and clearly respect human basic human rights and dignity. This requires the formulation of a human rights framework guiding the development of artificial intelligence based on cooperative consultation among various stakeholders, guiding and planning the future development direction of artificial intelligence to identify, prevent and correct the potential risks and harms that artificial intelligence may bring to basic human rights and dignity, and to safeguard human dignity and inherent value.

twenty one

. The application of educational artificial intelligence in education will likely replace many of teachers' daily tasks, but it cannot shake teachers' status as the main body that can independently establish teaching goals, and ignore the value and creativity of teachers' work itself, thereby reducing the status and dignity of teachers. Educational artificial intelligence provides learners with cognitive support and intelligent teaching support services, which should enable learners to make choices based on their needs and personality traits, rather than make choices on behalf of students, that is, they should fully respect the basic human rights and dignity of students as a person, rather than treat them as a certain target. Therefore, the creators, teachers and other staff of educational artificial intelligence should fully respect these human rights and dignity of teachers and students in an intelligent learning environment.

(V) Free and self-governance

The development of artificial intelligence requires a balance between human-led decisions and machine-led decisions, and should promote human autonomy and control rather than harming and undermining human freedom. In what way should humans choose to give AI what degree of decision-making power to achieve the goal of human self-select

twenty two

. Therefore, artificial intelligence should regard users as independent people, not objects of manipulation, and cannot deprive and influence human freedom and autonomy. Schools and institutions that deploy educational artificial intelligence have autonomy over what kind of social and cultural values ​​to pre-set. The creators of educational artificial intelligence cannot embed their own social and cultural values ​​into educational artificial intelligence, which will affect users' autonomy over culture and value systems. The teacher profession has its creative and autonomous side. Teachers have autonomy as to what teaching methods they choose, how to arrange teaching activities and conduct teaching evaluations. Students have the right to control their personal image, data and identity in educational artificial intelligence. They have subjective initiative in teaching support services and resource strategy recommendations provided by educational AI, and should not be punished in any form for failing to follow educational AI’s advice.

(VI) Responsibility and Accountability

When artificial intelligence has certain autonomy in behavior and decision-making, its responsibility and accountability mechanisms will become more complex. Therefore, a complete set of responsibility and accountability mechanisms should be established to ensure that artificial intelligence and its behavioral decision-making consequences can be held accountable, and how responsibilities should be shared among manufacturers, designers, operators and end users. A clear accountability mechanism needs to be based on system transparency, that is, a complete data tracking and recording plan should be established to keep the system algorithms and decision-making reasoning process transparent so that the responsible person can be quickly identified under any circumstances.

15

. Educational artificial intelligence may replace some of the teachers' teaching decisions and teaching behaviors, or provide them with advice on teaching strategies to support their decisions and behaviors. This responsible party and accountability mechanism for decision-making and behavioral consequences is relatively complex, which requires educational artificial intelligence to have certain foresight for its own support and intervention in education and teaching work. For the decision-making process of the algorithm, the process of teachers using educational artificial intelligence for teaching, and the behavioral data during students' learning process, educational artificial intelligence must have a complete record-keeping plan to provide a basis for the subsequent accountability process.

4. Conclusion

There is no doubt that artificial intelligence will have a profound impact on education and may even bring about a fundamental change in the field of educational and teaching practice, transforming the nature of teachers' work and social roles, and changing students' learning experience. The emergence of educational artificial intelligence has brought new challenges to educational technology. In the process of supporting educational and teaching practices, it reflects increasingly strong automation and subjective initiative, and will play the role of moral actors and recipients at the same time. It will complicate ethical issues in the field of education and raise a series of new ethical dilemmas. Therefore, before the large-scale design, development, deployment and application of educational artificial intelligence, it is necessary to seriously consider the construction of educational artificial intelligence ethical principles to lay a suitable ethical and moral foundation for the safe and effective use of educational artificial intelligence in the process of educational teaching practice.

This article was published in the 6th issue of "Electronic Education Research" in 2020. For reprinting, please contact the editorial department of the Electronic Education Research Magazine (official email:).

Please indicate the reference for citations: Deng Guomin, Li Mei. Discussion on ethical issues and ethical principles of educational artificial intelligence. Research on electronic education, 2020, 41 (6): 39-45.