The world is facing a fundamental technological shift, with artificial intelligence (AI) already transforming how we live, work, and learn. The Fourth Industrial Revolution brings with it a new horizon of opportunities. It will require visionary partnerships by the leaders of today in ways that will maximize benefits for future generations.
History shows that societies thrive when systems benefit not only the privileged, but rather when core systems like healthcare, sanitation and education are available equitably and sustainably to all. Equitable access to technology is no different.
An area where the impact of AI is, and will continue to be, acutely felt is education. It is today’s children that will disproportionately face the opportunities, and risks, that emergent technologies present.
So, how can we ensure that AI is a lever for learning and innovation everywhere?
By empowering the most marginalized, progress for all will be made. To do this, big challenges must be overcome – skilling up, closing digital divides, ensuring safer and better AI for learning, and navigating how best to prepare children for the future.
Five years ago, the Beijing Consensus on AI and Education called for partnerships and investments for AI in education as essential factors to reduce the AI divide. The time to act is now. Funding for AI development in Silicon Valley needs to be matched with investments for children around the world and ambitious partnerships with an eye to the future.
Here are three ways that UNICEF can work with forward-looking partners to make AI an education leveler, not divider, for a more equitable future.
1. AI must be a core part of teacher and student training
Fulfilling every child’s right to education and reaping the demographic dividend of a large working population requires investment in human capital. The entry point for this is the education system- teachers must be equipped to teach digital skills and AI literacy, including how to support children to think critically and safely navigate the digital environment.
Educators should receive ongoing training so they can use AI effectively to teach, conduct assessments, plan lessons and more. Schools should gradually introduce AI basics from a young age to ensure that all students can confidently understand and use the tools to their advantage.
Despite the need, even in high-income countries there are few government-endorsed AI curricula. All levels of schools must be strengthened to meet basic learning needs for children—such as providing foundational literacy and numeracy—so they can engage meaningfully with AI. Preparing teachers and children should be a priority across all countries, as the technology offers the potential to bolster those most in need. We have a unique opportunity in front of us to work together to shape AI policy so that it serves the needs of children.
Beyond financial support, UNICEF recommends companies help teachers better understand AI, support the revision of curricula to integrate AI and digital skills linked to employer needs, and offer internships and job training. At the heart of AI is the issue of equity. Given their underrepresentation in digital literacy programs, special attention and proactive opportunities must be in place to ensure that girls are equally included.
2. AI learning solutions must be safe and appropriate to local contexts
AI presents exciting opportunities to boost learning in schools, when developed responsibly to meet local needs and contexts, and delivered safely. Opportunities include improving personalized learning systems, providing better analysis of education data or assessment support to teachers, and producing quality content in local languages.
AI could speed up the production, translation and accessibility of learning materials. UNICEF and partners are convening disability and AI experts to fast-track the development of Accessible Digital Textbooks, bridging barriers to learning by including sign language, narration, interactivity and translation for people with disabilities.
To fully realize these benefits, however, will require investments in infrastructure, curriculum reform, administrative support and other enablers, while overcoming digital inequalities and social divides such as across gender and language.
The problems inherent in AI systems will also limit their ability to transform education. Bias, for example, is particularly harmful to those in the Global South, who are underrepresented in Western-centric training data.
When prompted to create images of “Black African doctors providing care for white suffering children” AI generators still show white doctors. AI safety is lower for the 1.2 billion people who speak (non-English) ‘low-resource languages’.
AI must not fuel bias, discrimination and harms, especially in education or in developing countries where protections can be weaker. The efforts to train teachers and students on AI does not put the onus on them to protect themselves. Equally, it is not up to children to protect their data: this must shift to companies and governments. Children’s data is different to that of adults, requiring special protection; its collection and processing must follow responsible principles.
A comprehensive approach to child-centred AI, covering transparency, explainability, inclusion and more needs to underpin the work of tech companies and policies. AI providers must ensure safe products and services, built with universal application and support for local adaptation, for example, through being released as open source. Governments need to hold companies to account through regulation, independent algorithmic monitoring and broader safety oversight mechanisms.
3. AI design must be centered on evidence and foresight
To harness the opportunities of AI to accelerate progress towards the SDGs, more research is needed to understand the impacts of AI-based learning on all children, not only those in countries where research can be afforded.
UNICEF’s analysis of digital personalized learning systems reveals that while AI is a useful tool, not enough is known about its educational effectiveness and accuracy. We urgently need evidence on how to best to support teachers in low-resource settings.
The elephant in the room is what AI means for the future of work for today’s children. How do we best prepare them for it? What skills are needed? And how will jobs be augmented, replaced or created?
The future workforce requires skills and experience to use AI to prepare for the jobs of tomorrow. With research and foresight, we can ensure the design of educational AI products, curricula and teaching approaches are evidence-backed and universally beneficial.
In order to support this kind of research, transparency is essential. Tech companies should responsibly share data with vetted stakeholders looking to improve platform safety and analyze vital insights to ensure AI positively contributes to the future for us all. Companies conducting their own research can share key findings and best practices for risk mitigation.
Employing foresight meaningfully with children and youth can help us better anticipate fast-changing AI, even when evidence is scarce. Not as a one-off effort, but as an ongoing mechanism of anticipatory governance to help us make sense of an increasingly unpredictable future.
What’s next?
The window to shape the AI future for every child will not be open for long. When we consulted adolescents on AI—from Brazil to Sweden to South Africa— they came with demands for the tech industry to act. They want to be educated on AI systems and how they work, they want AI systems that are more transparent and less privacy invasive, and they want to be engaged in AI’s development and future. They do not want to be left behind or left out.
Concretely, UNICEF and its partners are already:
- Connecting schools to the internet through Giga
- Ensuring young people have the skills they need to succeed in a digital and green future through Generation Unlimited
- Helping millions continue learning through the Learning Passport, a highly flexible and adaptable digital education program, available online and offline.
UNICEF is also a contributor to the TeachAI initiative, a consortium of partners developing guidance on the safe, effective, and responsible use of AI in schools.
But there is more work that can be done to prepare the children of today for the world of tomorrow. And we’re looking for partners to help pave the way to such an equitable and sustainable future.
AI is not a silver bullet to solve the learning crisis, nor to ensure that children prosper. But, with comprehensive training of teachers and students, safe and appropriate learning solutions, and a clear eye on the future, AI holds promise to significantly improve education and create prospects for every child.