Do we need to call for a more cautious approach to generative AI in education?

By Valtencir M. Mendes, UNESCO Senior Education Specialist, Chief of Education at the UNESCO Regional Bureau for Education in Latin America and the Caribbean.

According to Bloomberg Línea, ChatGPT, owned by the research laboratory OpenAI, is receiving 1.6 billion visits each month. This means that there will be about 16 billion visits to ChatGPT worldwide in the current year – a figure that is expected to grow exponentially. If we focus on Latin America and the Caribbean, there have been more than 300 million visits from the region so far this year.

In recent months, several generative artificial intelligence (AI) tools, like Google’s Bard chatbot or Microsoft’s Bing search engine have emerged. And now ChatGPT has launched a new feature with the ability to generate charts and maps and transform images into videos. There is no doubt that generative AI tools have the potential to be a gamechanger in several industries, including education.

As analysed in the 2023 GEM Report, education systems should always ensure that learners’ best interests are placed at the centre and that digital technologies are used to support an education based on human interaction rather than aiming at substituting it. But are education systems ready to do so, while generative AI is already disrupting the learning and teaching ecosystem?

To try to answer this question, let’s delve into the transformative world of generative AI, with a particular focus on ChatGPT. This tool demonstrates the ability to not only make predictions but also generate new content resembling human-produced material. However, it lacks real comprehension or awareness, it simply aggregates fragments of words using statistical probabilities to generate potentially valuable texts.

The development of AI applied to education dates back to the 1970s or even earlier, but it is in recent years that we have witnessed a remarkable surge in its implementation. The combination of big data, increased computational power, and internet connectivity has led to the exponential acceleration of AI systems, including machine learning and deep learning, which underpin ChatGPT.

Proponents of AI say that tools like ChatGPT will bring significant benefits. It can assist teachers, personalize learning with AI-powered tutoring systems and automate administrative tasks, allowing educators to focus more on teaching and mentoring students. However, current educational challenges in access, equity and inclusion in education can be significantly magnified by digital exclusion, lack of connectivity and the digital readiness of countries. On top of these challenges, we have many education systems still attached to teaching methods based on memorization, lack of relevance and engagement as well as poor and short-sighted standardized assessments that can hamper the effective integration of AI for an inclusive and quality education.

To fully leverage AI’s potential, we need to embrace a transformative education paradigm

A proper rethinking of what teaching and learning means is required that entails prioritizing critical thinking, collaboration and creativity-driven pedagogies, intercultural understanding and global citizenship education as well as AI literacy. If the education debate only focuses on how to integrate AI or develop AI skills, we will miss an opportunity to put students at the centre of the learning environment and advance what it means to be human, setting ourselves apart from machines. This requires enhancing education to ensure that students learn high-level thinking capabilities, ethics, values, subject and context-specific content as well as practical knowledge of AI.

Introducing AI in education is a complex matter and there is no one-size-fits-all approach. If we consider teacher adoption of AI tools, for instance, we might first need to show how AI can potentially reduce workload for educators and how it could be a tool for teacher’s creativity and ownership, as well as establishing teacher support systems, and addressing ethical concerns. Adequate teacher, educator and learner training is essential to integrate AI effectively and ethically into the learning environment.

Research has also identified other challenges in implementing AI in education. For instance, in order to ensure fair, inclusive, and equitable AI applications, we need to think how to integrate AI into the curriculum, develop ethical guidelines for AI in education, and address biases in data, among others. Additionally, ethical concerns regarding privacy and data usage require thorough consideration.

There are ethical concerns that cannot only be addressed at school level but require thoughtful regulation to ensure that AI serves human interests of inclusive education systems and avoids overreliance on technology companies. Only by preserving essential human intellectual abilities, while exploring AI’s capabilities, can we enrich our lives and walk towards an augmented human intelligence with the support of digital technologies, ensuring inclusive and equitable quality education and promoting lifelong learning opportunities for all.

However, we are still far from this scenario. As with any groundbreaking technology, ChatGPT is not without its controversies. Famously, thousands of individuals, including technology company founders, signed a letter demanding a pause in AI and ChatGPT research. Critiques from activists like Naomi Klein have also brought attention to ethical concerns and intellectual property issues.

The challenge confronting us is significant. UNESCO is at the forefront of supporting countries in this environment, and has been working deeply on the ethics of scientific knowledge and technology, including AI, for more than two decades. In 2019, the organization published the Beijing Consensus on AI and Education, providing guidance on how to leverage AI technologies to achieve the Education 2030 Agenda. As part of this consensus, UNESCO highlighted the need for ethical, transparent, and unbiased AI applications in education, recognizing the potential biases and dilemmas associated with data usage and privacy.

In November 2021, UNESCO produced the first-ever global standard on AI ethics – the ‘Recommendation on the Ethics of Artificial Intelligence’, a landmark framework adopted by all 193 Member States. The document highlights the importance of providing training and competencies in AI for educators and students. This includes AI literacy, basic arithmetic, and media and information literacy. UNESCO and other international agencies also encourage promoting women, people from diverse backgrounds, and minorities in the AI field to ensure inclusive innovations. Transparency, open data, interdisciplinary research, and critical awareness of AI’s limitations are crucial elements of the AI education curriculum.

The transformative potential of AI in enhancing learning experiences is recognized by academia and international organizations. However, we need more AI in education literacy, research on its impact, and a truly ethical AI implementation to ensure fairness, transparency, and respect for human rights.

The international education community’s commitment to fostering AI education should remain alert, creating a future where AI-powered learning promotes inclusivity, quality, and equity for all learners and an education centred around human interaction rather than seeking to replace it – keeping in mind Paulo Freire’s vision on the importance of a dialogical approach to education, where teachers and students engage in mutual learning and co-construct knowledge.

Frontier technologies are usually heralded and surrounded with promise, their advantages to the fore, with little consideration of the wider societal implications in their ethical and inclusive adoption or use. A cautious approach to the integration of AI in education is needed where further research, diverse multi stakeholders’ engagement and informed decision-making to navigate the complexities of an AI-driven world should be at the highest level of an inclusive education policy agenda.


Leave a Reply