Artificial intelligence is no longer a future concern. It’s already woven into everyday life. Search engines, recommendation systems, writing tools, grading software, hiring platforms, and increasingly, classrooms. As a result, calls for AI literacy have grown louder across education, industry, and policy.
But alongside those calls, another term keeps appearing: AI illiteracy.
At first glance, the pairing seems logical. If we can define literacy, then surely we can name its absence. Yet for those of us who have worked in literacy research for decades, this move should give us pause. We have been here before.
What Do We Mean by AI Literacy?
Across recent research, AI literacy is generally defined as a set of socio-technical competencies that enable people to understand, use, evaluate, and make judgments about AI systems. Importantly, this goes far beyond knowing how to “use a tool.”
Key components of AI literacy might include:
- Conceptual understanding: Grasping basic AI ideas (what AI is, how machine learning uses data)
- Practical skills: Knowing how to interact with AI tools (for example, how to write prompts or interpret AI outputs)
- Critical thinking: Evaluating AI outputs instead of accepting them blindly, including spotting errors or bias in AI-generated content
- Ethical and societal awareness: Understanding AI’s limitations and impacts (for example, recognizing that AI systems can inherit biases from their data)
- Adaptive mindset: Staying curious and updating one’s knowledge as AI technology evolves.
Taken together, AI literacy is not a checklist. It is a moving target , shaped by context, purpose, and technological change. What counts as competent engagement today may be insufficient tomorrow.
So Where Does “AI Illiteracy” Come In?
In much of the literature, AI illiteracy is defined implicitly as the absence of these competencies. Someone who does not understand how AI systems work, who over-trusts outputs, or who cannot evaluate AI-generated information is labeled “illiterate.”
This framing is tempting. It creates urgency. It signals risk. It gives institutions a problem to fix.
But it also imports a deficit model that literacy scholars have long critiqued. In simple terms, a deficit mindset is a way of thinking that focuses on what is missing, wrong, or broken rather than what is present, working, or possible.
Lessons From Earlier Literacy Debates
Those of us who studied online reading comprehension in the early 2000s will recognize the pattern. I started my research career in that space, where we worked to identify what “good” online readers did and what “bad” readers failed to do. What began as descriptive inquiry too often slid into prescriptive labels.
The result?
- Complex practices were flattened into skills lists
- Context and purpose were sidelined
- Learners were positioned as lacking, rather than developing
Over time, we learned to do better. To describe practices, conditions, and supports rather than sorting people into literate and illiterate categories.
AI literacy demands the same care.
Why “Illiteracy” Is a Problematic Label
There are at least three reasons to be cautious.
1. AI literacy is not stable
AI systems are changing faster than most educational frameworks can adapt. Labeling someone “AI-illiterate” today may say more about the speed of technological change than about their capacity to learn.
2. Context matters
A student, teacher, journalist, or clinician may demonstrate high AI literacy in one domain and limited literacy in another. Competence is situational, not universal.
3. Deficit labels distort responsibility
When we frame AI challenges as individual illiteracy, we risk obscuring the role of:
- Poor system design
- Opaque algorithms
- Institutional pressure to adopt tools without support
- Unequal access to training and time
In other words, people are not failing AI. Systems are often failing people.
A More Productive Way Forward
Rather than asking “Who is AI illiterate?” , better questions might be:
- What practices do people currently use when interacting with AI?
- What assumptions do they bring to those interactions?
- Where do breakdowns occur…and why?
- What kinds of instruction, time, and institutional support make responsible use possible?
Research already suggests that increased understanding of AI correlates with greater confidence, lower anxiety, and more thoughtful adoption. But that understanding develops through use, reflection, and critique, not labeling.
Building Literacy Without Building Deficits
If literacy is about participation in meaning-making practices, then AI literacy is about learning how to live and work well in AI-mediated environments. That includes knowing when to rely on AI, when to resist it, and when to refuse it entirely.
The goal is not to eradicate “AI illiteracy” as a condition. The goal is to cultivate:
- informed judgment
- ethical awareness
- adaptive expertise
Those are not traits people either have or lack. They are capacities that develop over time, in community, and with care.
You can follow my notes and thinking on AI illiteracy here.
In future posts, I’ll explore what this looks like in classrooms, professional settings, and everyday use. For now, this is an invitation to slow down the conversation. To be precise in our language, and generous in how we describe people learning to live with rapidly changing technologies.