AI Literacy Is Now the Law—Ignore It at Your Own Risk

The clock for AI Literacy is ticking. Why should you act now, what are the six pillars of AI Literacy, and how can you build on those?

Introduction

2023 was the year the world pivoted to AI. It was also the year i started developing the Pragmatic AI course. In the process, i have once more confirmed that by teaching others, we learn ourselves. Here’s what this journey taught me about AI, AI Literacy, AI courses, and actionable knowledge. With AI Literacy being a regulatory requirement as of February 2025, the time to act is now.

The first time i was asked to “teach AI” to an organization, the requirements were as vague as the expectations. AI hype was building up, and even organizations with a somewhat reluctant approach to innovation felt that they were risking falling behind if they did not get themselves up to speed with what everyone was talking about. It came down to this:

How do you make this intricate subject matter approachable to people who have zero technical background and very limited experience of using the technology itself, but at the same time have been exposed to a barrage of hype, misinformation and disinformation about AI?

How do you dispel myths, explain key concepts, share use cases, provide food for thought and teach hands-on skills in a half-day workshop, making it all approachable, digestible, balanced, and fun?

That forced me to not only think long and hard, but to invest in finding and preparing the right material. It was lots of work, but the interaction and feedback made it worth the while. Plus, this exercise advanced my own knowledge. I was forced to systematize my approach to AI, express it in a way the audience can relate to, and make it actionable.

The actionable part did not actually go far in that instance. We limited ourselves to criteria and techniques for choosing and using GenAI tools, which is what most people still equate AI with. The main goal of the course was to get enough background and exposure to AI to evaluate whether further investment was warranted. Mission accomplished.

The Business Case for AI Literacy

I was not specifically thinking about AI Literacy while preparing and delivering the initial Pragmatic AI course, and probably neither was the organization. We were focused on getting the job done, which meant providing learners the knowledge and skills they needed. For most people, the term AI Literacy started getting attention as a consequence of the EU AI Act.

Organizations tend to start thinking about regulatory compliance once the dust settles and the requirements have been clearly defined. The EU AI Act had not reached that point in 2023. The EU AI Act’s first provisions became effective as of 2 February 2025, and one of those is a legal requirement for AI Literacy.

Article 4 of the AI Act requires that ‘providers and deployers of AI systems shall take measures to ensure, to their best extent, a sufficient level of AI Literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training and the context the AI systems are to be used in, and considering the persons or groups of persons on whom the AI systems are to be used’.

In fulfilling this requirement, ‘AI Literacy’ means skills, knowledge and understanding that allow providers, deployers and affected persons – taking into account their respective rights and obligations in the context of the AI Act – to make an informed deployment of AI systems, as well as to gain awareness about the opportunities and risks of AI and possible harm it can cause (Article 3 sub 56).

That may sound flexible or vague, depending on how you look at it. The takeaways are that first, any organization building or deploying AI systems must ensure AI Literacy for their workforce, and second, what AI Literacy is exactly depends on the context. The EU AI Act’s mandate may not be directly enforceable yet, but the clock is ticking and it’s best to be ahead of the implementation timeline.

82% of leaders acknowledge that their employees need new capabilities to prepare for AI advancements, while 60% of workers admit they lack the skills to get their jobs done effectively. According to Gartner, “By 2027, more than half of chief data and analytics officers (CDAOs) will secure funding for data literacy and AI literacy programs”.

The Six Pillars of AI Literacy

If you are looking for a definitive framing of AI Literacy, reviewing literature makes sense. Almatrafi et.al studied AI Literacy conceptualization, constructs, and implementation and assessment efforts published between 2019 and 2023. As they note there are many different definitions of AI Literacy, depending on the domain or implementation level.

Even though definitions vary, the AI Literacy review identified six key constructs of AI Literacy across a total of 47 articles: Recognize, Know and Understand, Use and Apply, Evaluate, Create, and Navigate Ethically.

Recognition & Understanding

Recognize refers to the ability to differentiate between technological tools that utilize AI and those that do not. This comes down to the question “What is AI?”

Know & understand refers to the ability to understand AI fundamental concepts and techniques. This entails acquiring basic skills, knowledge, and concepts that do not require prior knowledge. For example, understanding how AI processes input data through machine learning techniques to arrive at the output.

Application & Implementation

Use & apply. This construct focuses on the operational aspect, specifically, the ability to use AI applications and tools and the ability to applyand integrate AI concepts to accomplish tasks. This is also related to the role of humans in the human-AI collaboration and inter­action, work-related capabilities, and the ability to adapt AI tools to achieve an objective.

Evaluation & Ethics

Evaluate. This involves the ability to analyze and interpret the outcomes of AI applications criti­cally. Having a comprehensive understanding of the technical aspects of AI enables individuals to examine and form informed opinions about their interactions with AI technologies.

Navigate ethically. An AI-literate person must be able to understand and judge ethical issues such as fairness, accountability, transparency, ethics, safety, privacy, employment, misinformation, ethical decision-making, diversity, and bias.

Development & Creation

Create. This construct emphasizes an individual’s ability to design and code AI applications. Some researchers claim that “create” does not correlate to AI Literacy and thus should be considered as a separate construct related to AI Literacy. But this point is really the most important one, as both research findings and experience in the field attest.

How Hands-On Learning Transforms AI Understanding

Upon reviewing different educational initiatives, some key findings AI emerged from the AI Literacy review. What stands out is the effect of project-based learning and developing applications. This approach has shown a significant, positive effect on the other dimensions of AI Literacy, namely, understanding, evaluating AI applications, and ethics. This is of paramount importance.

Building on the initial success of the Pragmatic AI course, I was asked to deliver it to more organizations. The course was delivered and evaluated by learners with different backgrounds, goals, and timelines. From corporations to NGOs, from 4 hours to 4 days, from managers to lawyers, creatives, entrepreneurs, support personnel, consultants, and executives.

Regardless of the context and setting, two points were consistent in my experience and present in all evaluations. First, the appreciation learners showed for the hands-on parts of the course. Second, the request to include more hands-on parts. There are numerous anecdotes i could share on the ways hands-on exposure advances AI Literacy.

Let’s consider the time when learners were asked to create their own AI model for image classification using a no-code tool. While many were struggling with accessing and evaluating the dataset, one of the most technically advanced learners was able to use the tool on a subset of the dataset.

Seeing how slow the training process was on his machine, and how big the dataset was, the learner came up with a suggestion: what if we could split the dataset among students? One student would process images of class A and train a model, another one images of class B, and so on. Discussing how that would lead to a multitude of models and figuring how to complete the exercise was an invaluable lesson.

From Understanding AI to Building with AI

With findings clearly emphasizing the importance of project-based learning approaches, you would expect these to be central for AI Literacy programs. However, this is far from true. The majority of the efforts studied in the AI Literacy review focused on “Know & Understand”. Then, in second place “Use & Apply,” “Evaluate,” and “Navigate ethically,” after that, “Recognize,” and finally, “Create”.

Experience confirms these research findings. In the last few months, I have embarked on a journey of discovery armed with access to some of the top online education platforms. The goal was to research and evaluate what is out there in terms of AI training programs. Different scope and context than the AI Literacy review, but some overlapping conclusions.

While the AI Literacy review included training aimed at various audiences, what i explored was a multitude of online training courses aimed exclusively at professionals. The courses ranged from entry-level to advanced. What I found here was a stark polarization.

Courses aimed at users without technical background were mostly variations of “Use ChatGPT for X”. Courses aimed at technical users were mostly variations of “Introduction to Y in Python”. Some of the technical courses were quite good. Most of the non-technical ones ranged from not very good to downright misleading.

What was the most striking finding of my research was the fact that non-technical users are not deemed capable of getting hands-on in anything except prompting large language models. My experience, however, shows otherwise. While trying to teach business users to code in Python may not make sense, there are ways to get them to practice the fundamentals that are both intuitive and useful.

That was also a key finding of the AI Literacy review. Programming knowledge does not appear to be a prerequisite to learning AI concepts. Children as young as three can understand AI concepts. Just because someone can’t code, it doesn’t mean they can’t think.

Why Current AI Training Fails Business Leaders

With an increasing number of AI project implementations, even seasoned professionals need to reorient themselves to the nuances of developing AI in order to manage and apply AI. Executives, managers and consultants are demographics of special interest, demonstrating a growing demand to go beyond understanding AI fundamentals. Entrepreneurs and creatives are adopting AI to boost their productivity.

The best way to learn is by doing, and this should be a guiding principle for all demographics. Developing an in-depth understanding of AI should be based on understanding and exploring different types of data, analytics, data science, data management and governance practices and tools. These are either missing entirely from AI educational courses for professionals, or covered partially in separate courses.

Equally impressive is the near absolute dominance of Generative AI in educational courses. Less than 10% of the courses I reviewed included modules focused on non-GenAI approaches. As for non-machine learning approaches, they were literally non-existent. There really is more to AI than this.

For coders and non-coders alike, most of the courses had one thing in common: over-reliance on specific solutions, most notably ChatGPT and the OpenAI API. That is somewhat understandable, but not necessarily wise.

Implementation and hands-on requires using specific solutions, and ChatGPT and the OpenAI API are probably the ones most people would recognize. But that doesn’t necessarily make them the best choice for educational courses, for a number of reasons. At least for coders there are alternatives and abstraction layers utilized in some courses, such as generic Python, PyTorch, and Keras.

Reviewing this wide array of courses provided valuable insights. Some courses were thoughtfully designed, with illustrative videos to explain the subject matter. Others simply gave access to a plain text document. Some courses had exercises (multiple choice tests typically) built in in the syllabus. Many coder-oriented courses also include lab modules, where learners are asked to complete specific tasks.

What none of the courses provided, however, is a holistic approach tailored to the needs of the most dynamic and creative professional demographics. An approach that goes beyond entry level recipes, piecemeal fragments, and technical jargon.

Bringing it all together

The combination of findings of the AI Literacy review and the ideas and experience I’ve gained by developing and managing AI projects, reviewing and taking AI courses, and the feedback I’ve gotten on the Pragmatic AI workshops is a solid foundation for a holistic course that serves learner needs.

The follow up to this article will elaborate further on AI Literacy and share an actionable plan to develop it by:

• Providing a skills assessment framework
• Outlining team development strategies
• Detailing a holistic AI Literacy program
• Sharing success stories

Subscribe to the Orchestrate all the Things newsletter and make sure you don’t miss out!

Soon, the full program for the first ever open on-site Pragmatic AI course will be published. Secure your 2025 AI Literacy training: Pre-register now for priority access to this best in class AI Literacy program. Limited seats available.