UNESCO: Call for Contributions: Definition of Algorithm Literacy and Data Literacy
Doug Belshaw1, Ian O’Byrne2, Tom Salmon3
1 We Are Open Co-op
2 School of Education, College of Charleston
3 Dept of Education, Rhodes University
v1.0 (6 July 2023)
In our digital world, the importance of literacy extends beyond the ability to read and write in a traditional sense. Not only is literacy also about participation, it is about power. Who gets to define what it means to read, write, and participate? What is the role of tech platforms?
As we navigate online spaces, engage with artificial intelligence (AI), and make decisions based on data, new forms of literacy emerge. Among these, AI Literacy stands out as a critical skill for the 21st century. AI Literacy is not just about understanding the technical aspects of AI, it's about being able to critically evaluate AI technologies, communicate effectively with AI, and use AI ethically as a tool in various contexts. This literacy is a cross-sectional 'slice' of the broader field of new literacies, intersecting with and complementing other forms of literacy such as algorithm and data literacies.
Our collective experience of efforts to describe what digital literacies and 'good AI' (ethical, responsible, healthy) look like[1] suggests that these should be theoretically grounded and stakeholder driven to ensure efforts shift norms and enrich lives.
"An audit-trail of decision-making is important, as it reveals both the explicit and implicit biases of those involved in the work, as well as lazy shortcuts they may have taken.” (Belshaw, 2016)
We do not see AI Literacy as a standalone concept, it intersects with and complements other forms of literacy such as algorithm and data literacies. There may be multiple, complementary definitions of AI Literacy, based on different understandings of what, for example, ‘algorithms’ or ‘data’ mean in a given context. Multiple definitions allow for multiple perspectives and voices, which is empowering in our diverse world.
UNESCO’s work on AI literacy and AI within education (AIED) speaks to these intersections, and should enable us to see how definitions came into being, the context in which they were developed, who was involved, and whose purpose(s) they serve.
AI Literacy is about more than just understanding the technical aspects of AI. It involves a set of competencies that enable individuals to critically evaluate AI technologies, communicate and collaborate effectively with AI, and use AI ethically as a tool in various contexts. These contexts are multiple and varied: home and the workplace, but also in social arenas and liminal spaces between all three. Consequently, definitions of AI literacy should speak to how AI is influencing our decisions, shaping our interactions, and challenging our notions of privacy and security in our daily lives.
Belshaw’s work (2016) highlights the cultural, cognitive, constructive, communicative, confident, creative, critical, and civic elements, offering a broad view of AI literacy. For example, this approach asks questions such as: How does AI influence our culture and cognition? How can we constructively use AI to solve problems or create new things? How can we communicate effectively with AI and use it confidently? How can we foster creativity through AI? How can we critically evaluate AI technologies and their impact on society? And lastly, how can we use AI in a way that promotes civic responsibility and contributes to the common good?
"These programs were never about terrorism: they're about economic spying, social control, and diplomatic manipulation. They're about power." (Snowden, 2014)
An algorithm is a set of step-by-step procedures, or a set of rules to follow, for completing a specific task or solving a particular problem. Algorithms power many different aspects of today’s society, from social networks and online shopping, through to credit scores and decisions around recruitment.
Data is a collection of facts, statistics, or information that can be qualitative (descriptive information) or quantitative (numerical information). It is used to inform many different decisions from healthcare to town planning, enabling governments stay up-to-date about their citizens, although this data collection by governments and large tech companies can also go too far[2].
Algorithm and Data Literacies are thus integral components of the broader digital literacy landscape, intersecting with and complementing AI Literacy, and are crucial for navigating the digital world effectively and ethically.
We understand ‘Algorithm Literacy’ as being aware of the use of algorithms in online applications, platforms, and services, understanding how they work, and having the ability to critically evaluate algorithmic decision-making. This includes the skills to cope with or even influence algorithmic operations that often operate behind the scenes, shaping our online experiences and influencing our decisions in ways we may not always realize.
Data Literacy, on the other hand, we conceptualize as the ability to read, write, and communicate data in context. It involves understanding data sources and constructs, applying analytical methods and techniques, and being able to describe the use-case application and resulting business value or outcome.
These literacies are not isolated but intersect with AI Literacy. Understanding how algorithms work (Algorithm Literacy) can inform our critical evaluation of AI technologies (AI Literacy). Similarly, being able to read and analyze data (Data Literacy) can enhance our ability to use AI as a tool for data-driven decision making (AI Literacy).
“AI relies on computational models, data, and frameworks that reflect existing bias, often resulting in biased or discriminatory outcomes, with outsized impact on marginalized communities.” (Mozilla, 2020)
A transformative approach to AI is essential for the future as highlighted by the EU Commission’s (2019) seven key requirements for ‘Trustworthy AI’ to ensure AI is lawful, ethical and robust. These include human agency and oversight; technical robustness and safety; privacy and data governance; transparency; diversity, non-discrimination, and fairness; societal and environmental well-being; and accountability.
Along similar lines, Mozilla, a pioneer in promoting an open and accessible internet, has done significant work identifying key challenges and guiding principles for developing and deploying AI systems. This research highlights that while AI has immense potential to improve our quality of life, its integration into everyday platforms and products can compromise our security, safety, and privacy. The challenges Mozilla (2020) identify include monopoly and centralization, data privacy and governance, bias and discrimination, accountability and transparency, industry norms, exploitation of workers and the environment, and safety and security.
The EU’s Ethics Guidelines hold open space for dialogue around privacy, transparency, and human well-being and rights as key considerations to ensure there can be accountability for harms. Similarly, Mozilla's guiding principles for AI - agency, accountability, privacy, fairness, and safety - provide a roadmap for developing more trustworthy AI systems that align closely with the Eight Elements of Digital Literacies (Belshaw, 2016).
By integrating these principles into AI Literacy, we can promote a more ethical and responsible use of AI. This involves understanding how AI works and advocating for AI systems that respect user privacy, provide transparency, ensure fairness, and contribute to a safer digital environment.
"Liberation is a praxis: the action and reflection of men and women upon their world in order to transform it." (Freire, 1970, p. 79)
AI, algorithms, and data are not just neutral technologies; they are embedded within economic systems and influenced by economic interests. This includes issues of monopoly and centralization, where a handful of tech giants dominate the AI landscape, stifling innovation and competition. It also includes issues of data privacy and governance, where people's data is often collected, stored, and shared in invasive ways for economic gain.
The transformational potential of AI is constrained by its hardest problems and runs the risk of deepening existing power inequalities. AI technologies struggle to address tasks that are not present in the training data, not always legible, or too high-stakes to deploy. Labor-intensive services like healthcare and education have proven hard to make more efficient. AI may be “transformative” by increasing productivity and changing habits, but as Mozilla’s work shows, if we ignore potential harms and implications for internet health and AI systems, these benefits may be undermined[3].
The expansion of AI knowledge and technologies into educational contexts via intelligent tutoring systems, assessment and evaluation, and adaptive learning systems has driven efforts to map out competencies and skills required for stakeholders in these areas. However, this focus on skills needs more ambition on ethics, it has paid less attention to issues of bias and meaningful AI transparency frameworks[4]. Transformative approaches, such as UNESCO's ‘AI and the Futures of Learning’ project need to be expanded. If we are to view literacy as a means of personal transformation and social change (Freire, 1970) we must advocate for a more emancipatory approach that empowers individuals and communities alongside economic interests.
"We must harness the power of the digital revolution to ensure quality education is provided as a public good and a human right, with a particular focus on the most marginalised" (UN Transforming Education Summit, 2022)[5]
The UN's Transforming Education Summit held in 2022 affirmed that teachers are central to the transformation of education. Similarly, literacies are not just about understanding and using technologies; they are also about transforming our world, promoting equity, and enabling human flourishing.
Equity is a fundamental principle that should underpin all forms of literacy. In the context of AI, algorithm and data literacies, this means ensuring that everyone, regardless of their background or circumstances, has the opportunity to develop these. It also means addressing biases and discrimination in AI systems, and advocating for more inclusive and equitable AI practices. Human flourishing requires that literacies should not just serve economic interests; they should also contribute to our well-being, creativity, and fulfillment.
Fundamentally, AI frameworks should be underpinned by a reimagining of our potential. The UNESCO Futures of Education lab (Sobe, 2021) has reworked each of the four pillars of learning (Delors, 1996) to build capacity for commoning actions and strengthening the common good. Similar work can help to further align UNESCO's frameworks for AI literacy with this transformative potential for education. Both Mozilla’s work and the Futures of Education work have highlighted the importance of addressing collective challenges, learning to collectively mobilize, learning to live in a common world, and learning to care and to inquire and co-construct the future together.
The challenges and opportunities of AI, algorithms, and data are not confined to any one country or region; they are global in scope. As such, our approach to these literacies should also be global, taking into account diverse perspectives and experiences, and promoting international collaboration and understanding.
AI, Algorithm, and Data Literacies are not just technical skills, but rather critical competencies for the 21st century. They intersect with and complement each other, reflecting the interconnectedness of our digital world. They require a humanistic approach, taking into account social, cultural, and ethical implications. And they demand a focus on equity, human flourishing, and the global commons.
UNESCO's work has always been grounded within a transformative approach to justice, equity, and human rights. UNESCO’s support for digital learning policies and AI literacy should remain grounded in robust ethical guidelines and regulatory frameworks. In support of this, we provide three recommendations:
[1] See Mozilla's Web Literacy Map, our published articles, and Doug Belshaw’s doctoral work on the Eight Elements of Digital Literacies
[2] As the revelations of Edward Snowden a decade ago showed (Guardian, n.d.)
[3] See Mozilla’s (2022) Internet health report
[4] See Mozilla’s (2023) work on meaningful AI transparency frameworks.
[5] See the UN Transforming Education Summit, held in New York in September 2022, which identified six actions, including “Digital Learning: Assuring quality public digital learning for all” - https://www.un.org/en/transforming-education-summit/digital-learning-all