“While GenAI should be used to serve education and research, we all need to be cognizant that GenAI might also change the established systems and their foundations in these domains. The transformation of education and research to be triggered by GenAI, if any, should be rigorously reviewed and steered by a human-centred approach,” United Nations Educational, Scientific and Cultural Organization (UNESCO) noted in its recently published ‘Guidance for use of Generative AI in education and research’.
Outlining a range of problems associated with the rising urge to use generative AI or GenAI for education, UNESCO has recommended a “whole-government” and multistakeholder approach for GenAI regulation in the education sector. Further, it primarily pitches for a human-centred approach as a guiding principle for developing regulatory frameworks and policies that will enable equitable use of AI in the context of education. These policies must mainly enable inclusive access to educational facilities, support personalised and open learning options, enhance the quality of education, monitor learning processes, and develop an understanding of the ethical use of AI.
Here, we summarise key sections of the guidance that not only cover broad-ranging issues but also delve into specific concerns that may impact teachers, learners, and individual researchers.
How to regulate the use of AI in education?
Based on the strategies explored by different countries since last November for regulating the use of GenAI, UNESCO recommended the following steps for regulation as well as effective use of AI in the education sector:
1. Legal framework for data protection, privacy
In order to tackle data protection and privacy risks, one of the primary challenges posed by generative AI applications, UNESCO recommends countries prioritise developing a legal framework for the collection and processing of personal data by generative AI services. This may be similar to the European Union’s General Data Protection Regulations (GDPR).
2. Implement national strategies on AI
According to UNESCO, by early 2023, some 67 countries had readied national strategies on AI. 61 of them could be considered standalone AI strategies, for example, India’s National Strategy on AI, while seven of the total were mainly chapters on AI as part of broader national ICT or digitalisation strategies. The guidance noted that none of these strategies cover generative AI as a specific issue. Hence, it is important that countries revise and expand the scope of their existing strategies.
3. Need for specific regulations on ethical use of AI
The guidance points out that countries have failed to identify ethical issues and specific safeguards required to prevent them in their AI strategies. Moreover, many countries have associated AI and education merely in terms of skill development, leaving out any discussion on domain-level ethical issues. UNESCO has urged countries to urgently articulate regulations on the ethics of AI and implement them. The guidance also suggests bringing in generative AI-specific regulatory frameworks built upon an assessment of gaps in existing local laws and regulations.
4. Revise or enforce Copyright laws
Unconsented or non-licensed use of Copyrighted works for AI training is another major problem that industry players and governments need to tackle. UNESCO urges countries to refine or adjust existing Copyright laws to govern the use of Copyrighted material for AI training as well as define the status of output generated by generative AI tools.
5. Capacity-building and reflecting on impact of AI on education
The paper suggests that countries need to facilitate capacity-building programmes for training educators in the proper use of generative AI and programmes for educational institutions to enable them to identify the potential benefits and risks of using AI. We elaborate on this below.
Article continues below ⬇, you might also want to read:
- Eight Key Points From China’s Interim Measures For Management Of Generative AI Services
- European Parliament Voted In Favour Of Adoption Of The AI Act
- Microsoft’s Blueprint For AI Governance In India: Testing High-Risk AI Systems, Licensing For Credibility, Promoting Transparency, And More
- Indian School Installs AI-Based Machine To Assess Food Quality For Students, What’s Concerning?
Guidance for policies concerning use of GenAI in education
1. Ways to promote inclusion, equity, linguistic and cultural diversity
The paper highlights that in order to tackle fundamental challenges in education, GenAI tools must be made inclusively accessible irrespective of gender, ethnicity, special education needs, socio-economic status, etc. These tools must advance equity, linguistic diversities and cultural pluralism by design. To achieve this, policy makers must:
- Identify the groups that cannot afford internet connectivity and take steps to bridge the digital divide to improve access to AI applications. This would also include establishing sustainable funding mechanisms for providing AI-enabled tools for learners with disabilities or special needs.
- Develop appropriate criteria for validating GenAI systems to prevent gender bias, discrimination against marginalised groups, or hate speech embedded in data or algorithms.
- Develop inclusive specifications for GenAI systems, such as data in multiple local and indigenous applications, and implement institutional measures to “protect linguistic and cultural diversities when deploying GenAI in education”. The paper adds, “Specifications and institutional measures should strictly prevent AI providers from any intentional or unintentional removal of minority languages or discrimination against speakers of indigenous languages, and require providers to stop systems promoting dominant languages or cultural norms.”
2. Protect human agency
As individuals increasingly use GenAI for assistance, one must be cautious of complete reliance on AI. As UNESCO noted, these tools must not be allowed to “usurp human thinking”. For this, policymakers must take measures to protect and enhance human agency while designing policies for education and research. They may consider the following points:
- Inform learners about the types of data that GenAI may collect from them, how these data are used, and the impact it may have on their education.
- Reinforce human control or choice in the context of using GenAI for teaching, learning, and research. Encourage critical questioning of norms or pedagogies that may be imposed using GenAI by educators, learners, and researchers.
- Prevent the use of GenAI where practical observations, empirical practices, social interaction, discussions, and logical reasoning are essential for learners to develop cognitive abilities and social skills.
- Prevent ceding human accountability to GenAI systems when making high-stakes decisions.
3. Monitoring GenAI systems for education
It is critical to monitor and validate GenAI systems for “ethical risks, pedagogical appropriateness and rigour, and its impact on students, teachers and classroom/school relationships” before deploying them for actual use. The guidance recommends following actions:
- Build validation mechanisms to test whether GenAI systems are free of biases and are trained on data representative of diversity (in terms of gender, disability, social and economic status, ethnic and cultural background, and geographic location).
- Address the complex issue of informed consent, particularly in contexts where children or other vulnerable learners are not capable of giving genuinely informed consent.
- Audit whether the outputs of GenAI include deep fake images, fake news, or hate speech. This would also require auditing such systems at the institutional level. Consider implementing minimum age restrictions for the independent use of GenAI in the institution.
- Adopt an “ethics-by-design” approach and exercise strict ethical validation of GenAI applications before they are officially adopted in educational institutions.
- Ensure that the GenAI apps in question are harmless, educationally effective, “valid for the ages and abilities of the target learners, and are aligned with sound pedagogical principles (i.e. based on the relevant domains of knowledge and the expected learning outcomes and development of values)”.
4. Developing government-endorsed AI-curricula in schools
According to the guidance, the development of “AI competencies” among learners is key to the safe, ethical and meaningful use of AI in education and beyond. This mainly refers to facilitating AI literacy covering both human and technological dimensions of AI, understanding how AI works and how it can impact stakeholders. The guidance recommends the following urgent actions:
- Commit to the provision of government-sanctioned AI curricula for school education, in Technical and Vocational Education and Training, as well as for lifelong learning. The curriculum must cover age-appropriate understanding of data and algorithms, impact of AI and the ethical issues it raises, and skills for creative use of GenAI applications, among other things.
- Support higher education and research institutions to enhance programmes to develop local AI talent.
- Devise methods for “intersectoral forecasts” for the impact of GenAI on the job sector at a national and global level.
- Enhance future-proof skills at all levels of education and lifelong learning systems based on prospective shifts in demand.
- Provide special programmes for older workers and citizens, who may need to learn new skills and adapt to new environments.
5. Adequate training resources for educators
Countries like China, Finland, Georgia, Qatar, Spain, Thailand, and Turkey, as per UNESCO, have developed or are developing frameworks for training programmes on AI for teachers. To build well-structured training programmes on AI for teachers, UNESCO recommends the following:
- Formulate locally relevant guidance for widely available GenAI tools and assist teachers in domain-specific uses of such applications.
- Define the value orientation, knowledge and skills that teachers need to understand and use GenAI systems effectively and ethically.
- Enable teachers to create specific GenAI-based tools to facilitate their professional development and learning in the classroom.
- Integrate emerging sets of values, understanding and skills on AI into the competency frameworks and programmes for training in-service and pre-service teachers.
6. Encourage diverse opinions and expressions
“GenAI, by definition, reproduces dominant worldviews in it[s] outputs and undermines minority and plural opinions. Accordingly, if human civilizations are to flourish, it is essential that we recognize that GenAI can never be an authoritative source of knowledge on whatever topic it engages with,” UNESCO noted. Hence, it is important to view output generated by GenAI critically.
To do this, UNESCO advises that stakeholders must consider the role of GenAI as a “fast but frequently unreliable source of information”. Moreover, so far, there is little solid evidence for determining the effectiveness of tools designed to support validated information. In such a scenario, it is important to encourage plural opinions and expressions of ideas to tackle homogenised and standard opinions. Learners must be provided with adequate opportunities to indulge in trial and error and empirical experiments to explore diverse observations.
7. Test locally relevant AI models and build a cumulative evidence base
The guidance noted that GenAI models are currently dominated by information from the Global North and under-representing voices from the Global South. Efforts such as harnessing synthetic data or computer-generated data are needed to make such models sensitive to the context and needs of local communities, particularly from the Global South. UNESCO recommends the following actions:
- Strategic designing and adoption of AI systems rather than indulging in non-critical procurement processes.
- Incentivize the designers of GenAI to target open-ended, exploratory and diverse learning options. ● Emphasis on tests and evidence-based use cases for applying AI in education in accordance with educational priorities rather than novelty, myth or hype.
- Use GenAI to trigger innovation and improvement in research methodologies through sophisticated technologies.
- Establish specific criteria based on evidenced pedagogical research and methodologies and build an evidence base for the effectiveness of GenAI.
- Strengthen evidence on the social and ethical impact of GenAI.
- Analyse the environmental costs of leveraging AI technologies at scale and develop sustainable targets to be met by AI providers to address issues related to climate change.
8. Reviewing impact of AI with an intersectoral approach
UNESCO has recommended policymakers to collaborate with AI providers, stakeholders in the education sector, and representatives of parents and students to plan “system-wide adjustments in curriculum frameworks and assessment methodologies” to explore the potentials of and mitigate the risks of GenAI.
Additionally, it is also advisable to involve intersectoral and interdisciplinary expertise from learning scientists, AI engineers, and representatives of other stakeholders to “examine the long-term implications of GenAI for learning and knowledge production, research and copyright, curriculum and assessment, and human collaboration and social dynamics”.
UNESCO’s guidance also provides detailed recommendations for facilitating “responsible and creative use” of generative AI for designing teaching and learning methodologies, curriculum framework, pedagogies for foundational learning, facilitating inquiry-based and project-based learning, and actively supporting learners with special needs.
Having said that, the paper also notes that as AI is bringing changes to teaching and learning processes, it is likely that the technology may undermine a learner’s direct engagement with educational content validated by humans. The paper highlights, “The authoritative appearance of GenAI text may mislead young learners who do not have sufficient prior knowledge to be able to recognize inaccuracies or to question it effectively. Whether learners’ engagement with unvalidated content should be recognized as ‘learning’ is also contestable.”
STAY ON TOP OF TECH POLICY: Our daily newsletter with the top story of the day from MediaNama, delivered to your inbox before 9 AM. Click here to sign up today!
