The Beijing Consensus on AI and Education and NITI Aayog’s Response

Niharika Ravi

Research Intern (former)

Indian Society of Artificial Intelligence & Law

Introductory Note

The outcome document of the International Conference on Artificial Intelligence and Education held in Beijing in May 2019, or the Beijing Consensus on AI and Education, was constructed with contributions from around 500 international representatives from over 100 member states, UN agencies, academic institutions, civil society and private sector members, and 50 government ministers and vice ministers in reaffirmation of the 2030 Agenda for Sustainable Development, and specifically SDG 4 i.e. ensuring inclusive and equitable quality education and promoting life-long learning opportunities for all.

Problem Addressed

Recent trends in AI, which push it to have profound effects on all walks of life, were recognised at the Beijing Conference and the potential to harness the benefits of AI to reshape the core principles of the teaching-learning process were addressed.


While the 2015 Qingdao Declaration committed to inculcating the use of Information and Communication Technology in Education in its commitment to SDG 4. The complexity of the rapidly developing AI technology was censured in the collective wisdom of the congregation in Beijing, pushing them to reaffirm UNESCO’s humanistic approach to the use of AI, and prioritising the protection of human rights ineffective human-machine collaborations for learning, sustainable development, and other goals.

The Consensus observed that AI development must be both humane and human-controlled, ethical and equitable, transparent and non-discriminatory, and took a strong stand for the impact of AI on society and people to be monitored throughout value chains.


These affirmations were connected to the recommendations made in Beijing by a set of common goals.

· Ensuring equitable access to education and AI

· Ensuring inclusive access to AI and education

· Ensuring life-long learning opportunities for all

· Avoiding the extension of social divide to digital divide in the use of AI in education

· Abiding by standards of ethics and transparency


Recommendations on AI and Education were made in keeping with the aforementioned affirmations and in tandem with SDG-4. The recommendations were made on the following broad fronts:

For Governments and Other Stakeholders in UNESCO’s member states

These recommendations are for the concerned parties to implement in accordance with their legislation, public policies and practices. Whole-government, inter-sectoral, and multi-stakeholder approaches to the planning and governance of AI in education was recommended to these parties with foundational planning on meeting local challenges to SDG-4 while being mindful of investment requirements on this front.

Considering using new models for delivering education to benefit all stakeholders is suggested but a censure is issued that human interaction between teachers and students must remain the core of education i.e. teachers cannot be replaced by machines. Governments are also advised to remain cognizant in AI’s potential to substantiate the learning process by transforming learning methodologies. These implementations are to be organised by drawing lessons from successful cases and scaling up evidence-based practices. In the same spirit, integration of AI-related skills in curricula is advised and the development of local AI talent is advocated.

The Beijing Consensus on AI and Education places significant stress on the fact that the development of AI education must not deepen the digital divide. In keeping with the letter and spirit of SDG-4, this policy pushes for technological advancement to facilitate learning anytime, anywhere, and potentially for anyone on a personalised level. The policy recommends paying attention to the needs of older people, and specifically older women on this front. On similar lines, ensuring AI provides high-quality education to all irrespective of gender, disability, socioeconomic status, ethnic or cultural background, or geographical location. Inclusion of those with learning impairments or studying in a language other than their mother tongue is proposed.

Ethical, transparent, and auditable use of education data and algorithms is proposed while it is propounded that the concerned bodies must be mindful of the lack of systematic studies of AI and education.

For International Organisations and Others Active in the Field

The congregation issued financing, partnership, and international cooperation guidelines to the concerned organisations to address the AI divide and disparity and giving focus to Africa and Less Developed Countries, etc. This portion of the proposal promotes collective action for equitable use of AI in education- global and regional. Alignment of international cooperation with national needs forwarded and in the context of the development of both, AI technology and AI professionals strengthens the recommendation to create multi-stakeholder partnerships to mobilise resources to bridge the AI and digital divide.

For the Director-General of UNESCO

The establishment of an ‘AI for Education’ platform is proposed to promote the use of AI under SDG-4 by providing a comprehensive database of open-source AI-related material tools, and policies among many other recommendations to further the international cooperation reinforced in this field by UNESCO by taking into account national and local needs.

Critical Appraisal

The Beijing Consensus on AI and Education makes a fair attempt to appeal to the growing need of regulating the use of artificial intelligence. Viewing that it looks exclusively at the applications of AI to education, the policy stands out as it proposes ideas for tackling challenges related to this field on local, national, and international fronts while prioritising the eradication of the digital divide- a hand-me-down of the previous era of ICT.

The acknowledgement that there is a dearth in academic research on the implications of introducing the use of AI in the education sector is reinforced by the focus on bridging the digital divide, for the use, and perhaps the misuse and inequitable use of AI in education may more likely be at the cost of the underprivileged on all fronts in society, as history has evidenced.

The Beijing Consensus was held in May 2019, but its applications to the Post-COVID era are stark in nature and must be addressed. The sudden shift from education in classrooms to that in virtual classrooms suddenly propelled the discussion on the use of machine learning and artificial intelligence in classes from either experimental IT or science fiction to immediate or potential reality. The multi-stakeholder approach proposed here, coupled with the multiple notes of caution on the use and misuse of AI in education, especially with regard to data security, AI ethics, and data privacy protection are matters that need immediate national and international attention.

NITI Aayog’s Response to AI and Education with focus on the Beijing Consensus

The NITI Aayog, a policy think tank established by the Government of India in 2015, introduced the national strategy for AI (#AIforAll) in 2018 in 2018, identifying 5 core areas to ensure AI progress in India in a discussion paper. The organisation’s CEO said that the paper lay the groundwork for evolving the National Strategy for AI and improved access and quality of education was notably one of the five areas mentioned in the paper.

The discussion paper identifies low retention rates and poor learning outcomes as issues that must be tackled on the Indian Education front along with multi-grade and multi-level classrooms, lack of interactive pedagogy, ineffective remedial instruction and attention for drop-outs, large teacher vacancies due to unequal concentration of teaching populations across the country, etc. as many problems that must be tackled. Low adoption of existing technology was notably a feature on the list even though EdTech is becoming a global phenomenon according to the paper. The paper proposes a two-pronged solution to these problems- introduction of adaptive learning tools for customised learning, intelligent and interactive tutoring system, automated personalization of teachers, and predictive tools to inform pre-emptive action for students predicted to drop out of school among others. This was before the Beijing Consensus.

To fill the gap in research illustrated by the Beijing Consensus, the NITI Aayog had also sought investment of Rs. 7500 crores to boost research and adoption of AI, with a high-level task force to oversee implementation in sectors including education, and to institute 5 research centres and 20 AI adoption centres. It was noted that the potential of such investment was an addition of $957 billion to the Indian GDP by 2035 and a boost of 1.3 percentage points in annual growth by the same year.

In February 2020, the think tank launched an AI module for school students in collaboration with NASSCOM in the form of Atal Tinkering Labs under the Atal Innovation Mission. The module comprises several videos, experiments, and activities that would teach students about the fundamentals of AI and prepare them for the digital era, in keeping with the provisions of the Beijing Consensus that advocate for AI education.

In July 2020, the organisation published a whitepaper on Responsible AI for All that can be viewed as a direct result of the Beijing Consensus’ stress on AI ethics, data privacy, AI security, etc. The white paper called for the introduction of Ethics in AI in mainstream university curriculums to encourage youth to explore unbiased and responsible uses of AI.

In November 2020, NITI Aayog proposed to set up an oversight body to play an enabling role across different aspects of AI and published a paper titled ‘Enforcement Mechanisms for Responsible #AIforAll.’ Research and education were among the spheres of influence that the body is set to have. This paper was instituted as the second portion of the aforementioned 2018 national strategy paper and comments from stakeholders were invited on both.

Concluding Observations

The Indian Government’s think tank NITI Aayog had already formulated a basis for its strategy on AI at a national level in 2018- prior to the Beijing Consensus. However, the actions taken by the think tank in the time subsequent to the Beijing Consensus have revealed that the NITI Aayog is, indeed, following the path laid down in the policy proposal formulated therein.

The absence of the NITI Aayog’s acknowledgement of the digital divide and the strategy to bridge it, especially after the Consensus’ stress on it is significantly glaring, especially in its 2020 publications wherein the pandemic had made the digital divide more eminent than ever before. On the other hand, the think tank has made significant strides in attempting to address research in EduTech, and specifically in the incorporation of AI and education in its many policies which is worthy of mention and praise in its commitment to the Beijing Consensus and to SDG-4- the foundation stone of this discourse on the incorporation of advancing technology in education.


The Beijing Consensus on Artificial Intelligence and Education

NITI Aayog, “National Strategy for Artificial Intelligence #AIforAll,” June 2018.

NITI Aayog, “Towards Responsible AI for All,” July 2020,

The Economic Times, “Niti Aayog proposes Rs 7,500 crore plan for Artificial Intelligence push,” May 20, 2019,

26 views0 comments