Recommendations on the Ethics of Artificial Intelligence The rapid advancement of artificial...
Report Interpretation | Guidance for generative AI in education and research from UNESCO
In the 21st century education landscape, technological innovations are rapidly transforming traditional teaching and learning methods like never before. The UNESCO Education 2030 Agenda advocates for the ambitious goal of "ensuring inclusive and equitable quality education and promoting lifelong learning opportunities for all." With the continuous advancement of artificial intelligence technology, generative artificial intelligence (GenAI) has emerged as a novel tool, showing its unique potential and value in the field of education.
Image from UNESCO
The "Guidance for Generative AI in Education and Research" (hereinafter referred to as "the Guidance") is the inaugural normative document issued since the inception of ChatGPT and the achievement of over one million users. It serves as a guiding document for regulating content and behaviors related to generative artificial intelligence. The guidelines emphasize manual review, labeling, and ethical considerations of generated content. They analyze the potential explicit and implicit risks of Generative Artificial Intelligence (GenAI) from six aspects: generated content, policy and regulations, intellectual property rights, and digital divide. Furthermore, specific measures to mitigate these risks are proposed from the perspectives of governments, service providers, and users.
UNESCO Global Guidelines for GenAI
Image from UNESCO
As a global leader in education and cultural affairs, UNESCO has issued a series of global guidelines for the application of Generative Artificial Intelligence (GenAI) in the field of education, aimed at ensuring that these new technologies have a positive impact on education while protecting core human values.
Explicit Risks of Generative Artificial Intelligence
Image from MulitIBrief
Generative artificial intelligence (GenAI) is not a novel concept, but before its widespread application in daily work and learning, it is crucial to clarify the potential risks associated with this technology.
UNESCO emphasizes the need for immediate action by countries to address the challenges posed by GenAI and to develop long-term policies to adapt to the rapid development of technology. This includes assessing the potential risks of GenAI tools and implementing corresponding regulatory measures.
Disconnection from reality leads to trust deficit. GenAI models do not output results through observation of the real world or scientific evidence, thus the accuracy and authenticity of the generated text are questionable. Moreover, these intelligent models are not built upon a genuine understanding of language and real-world societal values, often deviating from human-oriented perspectives and potentially producing misleading terminologies and discourses. If teachers and students reading and using such content fail to think critically and skeptically and instead choose to rely on and trust it, there will inevitably be risks to the transmission of knowledge, skills, and values, leading to a deficit of trust.
Image from HillPost
Lack of specific regulations leads to weak supervision. Technological legislation often lags behind the pace of technological development, lacking guidance and regulation in terms of laws and regulations. Many companies increasingly using GenAI services have found significant challenges to the security of their systems. Although the application of artificial intelligence can significantly enhance people's ability to complete tasks, the government's supervision and regulation policies and methods for many companies providing GenAI services are limited, leading to the proliferation of arbitrary data acquisition and usage. Appropriate legislation is needed to protect the rights of individuals and institutions. The "Guidance" points out that China, EU countries, and the United States have introduced policies and regulations and revised specific laws to regulate the risks associated with Generative Artificial Intelligence.
Image from Appen
Unauthorized data sources infringe intellectual property rights. It is well known that Generative Artificial Intelligence models are built on a large amount of data, much of which is scraped from the internet, often without the authorization of the data and information owners. If unauthorized data is used to generate text, images, etc., it may be accused of infringing intellectual property rights. Researchers, teachers, and learners need to not only understand the rights of data owners but also raise awareness of intellectual property protection when using Generative Artificial Intelligence.
Image from LinkedIn
Blurring the line between reality and simulation leads to the spread of false information. Due to the lack of strict regulations and effective supervision mechanisms, the text, images, and videos generated by Generative Artificial Intelligence applications are often difficult to distinguish between real and fake. In a sense, Generative Artificial Intelligence models make deepfakes easier and more accessible, and the cost of producing fake news will be lower. For young people whose worldview and values are still forming, if they lack solid, systematic knowledge and the ability to discern the authenticity of information, Generative Artificial Intelligence tools and applications can pose significant hazards and risks to their development. Therefore, researchers, teachers, and learners in the education sector need to enhance their ability to distinguish between true and false information in Generative Artificial Intelligence output materials.
Image from security.org
Implicit Risks of Generative Artificial Intelligence
While Generative Artificial Intelligence (GenAI) brings explicit risks, due to the limitations of the technology itself, it also deepens risks such as information cocoons and digital divides, exacerbating social inequality.
Marginalization of vulnerable groups leads to uneven development. The data source of Generative Artificial Intelligence models is the internet. If a particular topic or content frequently appears on the internet, it will be transmitted to the model, which tends to repeat these topics and information in its output results. This may limit and undermine the effective expression of diverse opinions and pluralistic thinking. In regions lacking data, online content about the population in those areas is relatively limited, and their voices are often unheard or drowned in a sea of data, leading to further implicit risks of marginalization. Therefore, the "Guidance" clearly recommends that policymakers in each country should be aware of and take action to address the exacerbation of inequality among different groups and regions due to the widening gap in the use, training, and control of general models.
Image from security.org
Generative Artificial Intelligence (GenAI) exacerbates poverty gaps and deepens the digital divide. As an emerging AI technology, GenAI represents an iterative innovation of traditional AI architecture, relying on vast amounts of data and immense computational power. Consequently, GenAI can only proliferate in technologically advanced countries and regions and is controlled by a few economic entities. This widening digital poverty gap further marginalizes technologically disadvantaged regions, leading to the neglect of their voices. The potential risk is the widespread dissemination of the values of developed economies, exacerbating the digital divide irreversibly.
Measures to Mitigate Risks Associated with Generative Artificial Intelligence (GenAI)
The emergence and development of Generative Artificial Intelligence (GenAI) bring infinite possibilities to human production and life, particularly in education. Effectively preventing the risks associated with GenAI can help us fully harness its advantages and better apply it to life and learning scenarios.
Image from LinkedIn
1.Strengthen External Supervision and Regulation: Enhancing external supervision and regulation is crucial for mitigating risks associated with GenAI. Government regulatory agencies should guide and manage the design and evolution of GenAI through policy legislation. Policy frameworks should align with the legislative and regulatory backgrounds of various countries and be updated with the evolution of AI technology to enhance the applicability of existing regulations. While regulating, it's essential to provide AI innovation with the freedom to collaboratively create higher-quality content.
2.Foster Responsibility among GenAI Providers: Companies providing GenAI should adopt responsible and ethical attitudes, offering universal systems and applications to meet the needs of teaching, learning, and research. Firstly, GenAI providers should ensure the credibility of their models and data sources and comply with relevant intellectual property laws. Secondly, there should be robust oversight of GenAI outputs to prevent discriminatory and biased content. Finally, GenAI providers need to provide legal explanations for the opaque internal operations of their models, ensuring transparency.
3.Standardize Internal Assessments for Institutional Users: Educational institutions, as users of AI, need to regulate and regularly audit GenAI tools, protecting user data and automatically filtering inappropriate content. They should also assess and address the long-term effects of GenAI application in the education sector on student creativity and critical thinking. Additionally, institutions should set minimum age restrictions for independent users, such as recommending a minimum age of 13 for using ChatGPT, to ensure responsible AI usage.
4.Encourage Self-Restraint among Individual Users: GenAI tools are currently used by various groups, including researchers, teachers, and students. However, the process from emergence to widespread use is driven by technology rather than demand, lacking consideration for ethical principles. Users should enhance their self-restraint awareness and actions, promptly reporting any illegal GenAI programs to regulatory agencies to ensure lawful and compliant AI use in society.
Challenges and Opportunities in Implementing GenAI
Despite the enormous potential of Generative Artificial Intelligence (GenAI) technology in the field of education, its prospects for enhancing learning efficiency, enabling personalized instruction, promoting cross-cultural communication, and assisting special education needs are promising. However, at the same time, the challenges faced in the implementation process are equally significant and cannot be overlooked. These challenges not only involve issues of unequal access at the technical level but also encompass new demands on teachers' professional capabilities, as well as the complexities of ethical and legal issues.
Unequal Access to Technology
The introduction of GenAI systems may exacerbate existing disparities in access to technology and educational resources. While these tools have the potential to revolutionize education, not all students have equal opportunities to use them. Certain groups may face challenges due to factors such as economic conditions, geographical location, or family background, which may result in insufficient hardware devices, unstable internet connections, or inability to afford related service fees, thus preventing them from fully benefiting from the learning convenience and resource richness brought about by GenAI.
Deepening of this technological divide undoubtedly exacerbates issues of educational equity, contradicting the goal of ensuring inclusive, equitable, and quality education set forth in the global Education 2030 Agenda. Therefore, education policymakers and implementers need to actively seek solutions, such as providing public access facilities, subsidizing low-income families, and optimizing network infrastructure in remote areas, to ensure that all students can overcome the digital divide and participate equally in the learning environment empowered by GenAI.
Image from Washington Post
Teacher Training Needs
The widespread application of GenAI poses new demands on teachers' professional knowledge and skills. On one hand, teachers need to have sufficient digital literacy and AI knowledge to effectively assess and select appropriate GenAI tools, integrate them into teaching practices, and identify and correct errors in AI translation and interpretation to avoid cultural conflicts caused by misunderstandings. On the other hand, teachers also need to know how to guide students to critically evaluate the content generated by GenAI, understand its limitations, such as potential biases and data incompleteness, and cultivate students' autonomy and innovation capabilities, rather than relying excessively on the preset answers provided by GenAI.
This implies that the teacher training system must keep pace with the times, incorporating courses and training programs related to GenAI to help teachers master the operation and application strategies of GenAI tools proficiently, while also enhancing their legal awareness in ethics, data privacy, intellectual property rights, and other areas, ensuring responsible and creative use of GenAI in educational activities.
Image from Leverage Edu
Ethical and Legal Issues
The application of GenAI in the field of education raises numerous ethical and legal considerations.From an ethical perspective, GenAI may reduce direct interaction among students, affecting the cultivation of social and emotional skills, while excessive reliance on AI assistance may weaken students' autonomy and innovative thinking. Additionally, protecting student privacy, respecting intellectual property rights, and preventing inappropriate content generated by AI, such as deepfake images, fake news, and hate speech, become important tasks for educational institutions.From a legal perspective, defining the responsibility of students using AI-generated text, especially in cases involving new forms of academic integrity breaches, requires clear policy guidance and effective regulation.
Therefore, educational and research institutions need to establish strict ethical frameworks, conduct pre-screening, ensure that GenAI tools align with educational purposes, and are consistent with student interests. In the short term, strengthening manual detection methods to maintain academic integrity is necessary; in the long term, the assessment methods for assignments should be redesigned to focus on assessing students' unique human values when facing complex issues, rather than solely relying on GenAI.
Image from Medium
Thus, only by doing so can we ensure that the essence and value of education are not eroded while enjoying the convenience brought by GenAI.
In conclusion, although GenAI technology brings unprecedented opportunities to education, it requires addressing multiple challenges in practical applications such as technological access inequality, teacher training needs, and ethical and legal issues. Only through continuous policy guidance, resource investment, teacher training, and institutional innovation can we ensure that the application of GenAI in education truly serves the global education agenda of 2030, promoting inclusive, equitable, and high-quality lifelong learning opportunities and driving education towards a more intelligent and humanized direction.
If you are interested in obtaining the full version of the 'The Guidance for Generative AI in Education and Research,' please email the Center for Teaching and Learning at ctl@wku.edu.cn
Resources:
1.Institute of Smart Learning, Beijing Normal University. (2024). Compilation of materials for the 2024 World Digital Education Conference. Retrieved from Ministry of Education website, World Digital Education Conference website, & Micro-Expression Education.
2.Ministry of Education: World Digital Education Conference- Digital Education: Application, Sharing and Innovation http://en.moe.gov.cn/features/2024WorldDigitalEducationConference/#ShanghaiInitiative
3."World Digital Education Conference", https://wdec.smartedu.cn/en/
4."World Digital Education Conference concludes with multiple achievements",https://www.chinadaily.com.cn/a/202401/31/WS65ba2e78a3104efcbdae8f84.html
5.“Cooperation initiative agreed as conference on digital education closes”, https://www.shine.cn/news/metro/2401316635/
Author: Yirui (Sandy) Jiang
Chief Editor: Yirui (Sandy) Jiang