AI literacy is one of the first provisions of the EU AI Act to enter into force. It therefore warrants early consideration and planning. After an organization has assessed its AI literacy needs, the next step is to design a comprehensive AI literacy program. The goal is to meet legal obligations and foster a culture of responsible AI use that supports compliance, innovation and commercial strategy.
Why a structured AI literacy program is necessary
Organizations must ensure their staff and other individuals involved in the production, deployment or use of AI systems possess a sufficient level of AI literacy. This literacy must be proportional to their role, technical knowledge and the context in which the AI systems are being produced or deployed. Without a well-designed program, AI literacy efforts risk being fragmented, insufficient or misaligned with the organization's actual needs.
A systematic approach enables organizations to manage AI-related risks through consistent training standards, achieve scalability in delivery and track outcomes for compliance purposes. Moreover, a well-structured program allows for better integration with existing compliance efforts while maintaining cost-effectiveness.
Step 1: Define learning objectives and outcomes
Once the initial AI literacy assessment has been completed, you should have a good idea of the audience and their general needs, the next step is clearly defining the learning objectives for different groups/roles. For those under the EU regime, learning outcomes must align with the core competencies required under Article 3(56) of the EU AI Act, which defines AI literacy in terms of the skills, knowledge and understanding necessary for responsible AI deployment. While applicable in most scenarios, those outside the European regime have some flexibility and may opt for different learning objectives.
The process of defining appropriate learning objectives is likely to involve a combination of formal assessments such as via AI risk assessments, data protection impact assessments or general assessments of minimum corporate knowledge standards, and consultations with domain experts and regular AI users. Typically, it will be the senior staff from different teams that are best placed to specify the learning objectives of their specific teams/roles. For example, senior technical staff are best placed to assess the competencies required within their teams, provided of course, that they have a sound understanding of the baseline legal requirements.
When senior staff cannot specify AI literacy needs within their team, this is in itself an indication of a training need or a lack of strategic direction on the organization's use of AI. In either case, early alignment with senior stakeholders will be important to ensure everyone is pulling in the same direction and to identify areas where input by outside consultants in specific domains may be required.
Learning objectives should be tied to metrics from the outset so they can easily be assessed against effectiveness measures once the program is delivered.
Key takeaway: This is a good topic for discussion at AI governance forums for organizations that have them. Ideally, stakeholders from different teams will contribute to the discussion and can form part of a regular review process by having AI literacy as a regular agenda item. A common model is to provide these stakeholders with enhanced training to guide their respective teams; a "train-the-trainer" approach. If not already in place, consideration should be given to whether this mechanism may be a valuable addition to your AI literacy program.
Step 2: Plan content and delivery methods
Content planning should focus on addressing the identified learning objectives efficiently and effectively. The key is purchasing and/or developing material that resonates with different learning styles and organizational contexts while maintaining consistency in core messages about the organizational approach to managing AI.
Multimodal learning is an approach that utilizes various methods of conveying information. On a spectrum of most standardized to highly bespoke solutions, these options may include the following.
Generic corporate training videos
This method is best for general learning needs to be delivered at scale, as it requires little customization. Different off-the-shelf solutions and providers could form part of the organization's learning management system and be administered as part of a regular training cycle. This mechanism allows for generalized assessments at the end of the training modules to check for a basic level of understanding.
Workshops and seminars
Best for interactive learning, workshops and seminars can be conducted in-person or virtually and can engage employees with real-time problem-solving, case studies and collaborative discussions. This method allows for practical approaches and detailed discussion about considerations specific to the organization, industry or product.
Mentorship and peer learning
Pairing employees with more AI-literate colleagues can facilitate knowledge transfer and provide practical, hands-on learning experiences. This method allows for more personalized and bespoke training designed with the individual and their personal development needs and goals in mind. It also enables servicing specific needs, such as dyslexia, ADHD, autism or any neurodivergent needs of the individual.
Bespoke corporate training videos
This method is best for catering to the diverse needs of different employee groups. For example, technical teams might have in-depth training on ethical AI design, while customer services teams may focus more on how the AI system works, risks to customers, and how to resolve escalations or issues
Simulations and use-case scenarios
For high-risk scenarios, simulations and use-case scenarios can help employees practice responding to real-world AI challenges. This method is likely to be the most expensive but will be able to deliver practical and highly bespoke organization- or product-specific training.
Industry peer learning groups or initiatives
While the legislation is yet to be fully enacted, it may be useful to work with industry groups and trade associations or similar bodies to understand where your organization fits with what peer companies are doing to benefit from industry insights and initiatives.
The role of compliance professionals is to make sure the content of the AI literacy training program meets the legal requirements, matches the needs of identified roles and risk profiles, and meets the learning objectives.
It is worth setting aside some time to explore modern solutions beyond the traditional corporate training providers. For example, modern behavior focused e-learning platforms are very well suited to multimodal learning experiences and are able to support tailored training modules, which allow the content of the training to be more specific to the group or team it is being delivered to.
Integration with platforms also offer interesting opportunities such as the ability to create text-to-video content. This allows for easy updates and customization, an aspect traditional corporate video providers have long struggled with and which will become increasingly important in rapidly evolving and customization-heavy scenarios like AI literacy.
Step 3: Allocate resources and set a timeline
For AI literacy programs to be successful, organizations must allocate sufficient resources, including budget, time and human capital. This involves not only setting aside funds for external training providers or online platforms but also ensuring employees have the time and support they need to complete the training.
However, resources are often limited, particularly in compliance and governance teams, therefore prioritization of learning needs against both risks and organizational goals is likely necessary. It is helpful to document desired learning outcomes even when these exceed available resources as these gaps can be used to inform program development, and potentially help teams identify needs that may be filled by recruitment.
The timeline for rolling out an AI literacy program should be realistic, accounting for the complexity of the training and the availability of key employees. In many cases, a phased approach in which high-priority groups such as those working with high-risk AI systems receive training first, followed by other departments, works best. Continuous development and regular updates are essential as AI technologies and regulatory requirements evolve over time.
Step 4: Measure success and adapt the program
An effective AI literacy program requires mechanisms to track and measure success. As AI literacy is outcome focused, i.e., ensuring sufficient skills, knowledge and understanding, it cannot be measured in terms of training provided but must be understood in the context of achieving these objectives. This can be measured through:
- Assessments and quizzes: Testing employees' understanding of AI concepts, risks and regulations at the end of each training module.
- Feedback mechanisms: Collecting feedback from employees on the relevance and effectiveness of the training, as well as any challenges they encountered in applying their new knowledge.
- Performance metrics: Tracking key performance indicators such as the reduction of AI-related errors or complaints, improved compliance with AI governance policies, and incident related metrics.
Once in place, an AI literacy program should remain adaptable to address new challenges and opportunities. This involves regularly updating the training content as AI technologies evolve, as well as incorporating lessons learned from incidents, audits and feedback.
Conclusion
Designing a comprehensive AI literacy program involves translating identified needs to measurable learning objectives, content planning, delivery methods, resource allocation and success measurement. By taking a structured approach to these elements, organizations can build programs that not only meet compliance requirements but also drive meaningful engagement with AI systems across their workforce.
Regular review and adaptation of the program ensures it remains aligned with both organizational needs and evolving regulatory requirements, supporting long-term success of AI governance and workforce capacity building.
Erica Werneman Root, CIPP/E, CIPM, is the co-founder of Knowledge Bridge AI and a consultant via EWR Consulting.
Monica Mahay, CIPP/E, CIPM, FIP, is the director of Mahay Consulting Services and chief compliance officer at Sky Showtime.
Hatla Færch Johnsen is the CEO and co-founder of uQualio Video4Learning.
