THE Artificial Intelligence (AI) has become a constant presence in discussions about the future of education. Examples of virtual assistants, personalized learning experiences, and new models of interaction between students, teachers, and digital platforms are multiplying.
However, there is an interesting contrast between the enthusiasm generated by the technology and the evidence currently available. One of the most relevant conclusions of Open edX Conference 2026 It was precisely the observation that, despite a significant portion of the presentations addressing AI, Few projects have shown robust results measured in real-world learning contexts.This doesn't mean that AI has no potential, it just means that we are still in a phase where experimentation is more abundant than evidence.
An interesting example was a study with approximately 20,000 participants that compared summary videos presented by humans and by AI-generated avatars. The results did not identify significant differences in the indicators analyzed. However, the researchers themselves cautioned against a hasty interpretation of these results: they were only evaluating short, supplementary content and not core learning activities, complex problem-solving, or topics with a higher emotional charge. The recommendation was not to abandon this technology, but to continue experimenting, measuring, and producing evidence.
It is precisely this idea of responsible experimentation that seems most relevant to us in the context of online education. At NAU, the FCT's digital service, developed by FCCN, we believe that the question is not whether Artificial Intelligence will be part of the future of digital learning. The question is to identify where it effectively creates value for students, teachers, pedagogical teams, and partner entities.
In an ideal scenario, it would be possible to provide intelligent assistants to all participants in all courses. However, reality requires considering factors such as usage costs, financial sustainability, data protection, and the effective impact of the implemented solutions. The Open edX community itself (of which NAU is a part) acknowledges that many doubts still exist regarding the behavior, scalability, and costs of these models when used on a large scale.
Therefore, it makes sense to start with the areas where the potential return is highest. In our experience, this value is often found in supporting content production, assisting course teams, analyzing participant feedback, monitoring pedagogical quality, and improving platform management processes.
Experience shows that AI is particularly effective as a tool to support the creation and review of content. However, the value of these processes continues to depend on human intervention. In an educational platform, knowledge cannot be treated as an automatically generated product. The expert in the field continues to play an irreplaceable role in scientific validation, critical framing of content, and ensuring its pedagogical quality. AI can increase productivity, but the credibility of knowledge continues to rest on human experience, rigor, and judgment.
This perspective also applies to the use of AI in pedagogical processes. Replacing teachers, authors, or specialists should not be the goal. The aim of this use should focus on increasing the capacity of these stakeholders to create better learning experiences, freeing up time for higher value-added activities, such as student support, pedagogical design, or ensuring the scientific quality of the content.
At NAU, this approach is already translating into concrete initiatives. One of the areas where we have been exploring the use of AI is in the initial phases of defining the scope of courses and their respective course design. The possibility of using these tools to support the structuring of content, identify gaps, organize learning paths, or generate activity proposals allows us to accelerate processes that naturally remain dependent on the validation and decision of the experts responsible for each course.
There are also already cases of this use on the platform. In the context of content production, some projects funded by the Recovery and Resilience Plan have allowed the exploration of the use of AI technologies in the creation of educational resources, namely in the production of videos. These experiences have been important for better understanding not only the potential of these tools, but also their limitations and the contexts in which they can effectively add value.
At the same time, we closely monitor the technological evolution of the Open edX platform itself, the open-source software "engine" of NAU. We have already begun installing and experimenting with the functionalities made available by the community through Open edX AI Extensions, an initiative that seeks to create a common basis for integrating AI capabilities into the Open edX ecosystem. The Open edX 2026 conference itself demonstrated that this area is still in a relatively early stage of maturation, with important issues remaining related to scalability, usage models, costs, and evaluation of pedagogical impact.
It is precisely because we recognize that many questions remain unanswered that we actively participate in the Artificial Intelligence working group of the Open edX community. More than just following trends, we seek to contribute to the discussion about the paths the platform should take in this area, bringing to the table the perspective of a large-scale public online education infrastructure that must balance innovation, sustainability, pedagogical quality, and responsibility in the use of available resources.
Perhaps the best analogy is that of an investor. There are times when it's worthwhile to take risks to explore new opportunities. There are others when it's important to consolidate knowledge and realize the effective return on investments made. A startup can adopt a stance closer to that of a bold investor, experimenting rapidly and accepting failures as part of the process. A public education platform has a different responsibility. It's not about avoiding innovation, but about ensuring that each investment generates demonstrable value for the community it serves.
This doesn't necessarily lead to a conservative stance. On the contrary, it requires a more demanding approach: testing, measuring, learning, and scaling only what demonstrates results.
At a time when AI seems to offer solutions to almost every problem, perhaps the most important question is not what the technology can actually do. Perhaps we need to understand, based on data, where it truly makes sense to use it. Online education doesn't need more technology for technology's sake. You need solutions that contribute to more effective learning....more efficient management and a greater capacity to respond to the needs of its users. And it is precisely in this balance between innovation, evidence and responsibility that Artificial Intelligence can play a transformative role.
Transparency note
This article was developed with the support of generative Artificial Intelligence tools. AI was used as a mechanism to support the analysis of the Open edX 2026 – The Full Breakdown report, information synthesis, argument structuring, and linguistic revision of the text. The ideas, positions, interpretations, and strategic framing reflect the author's vision on the adoption of Artificial Intelligence in online education and were fully reviewed and validated by human intervention. The use of AI did not replace critical thinking, source analysis, or authorial responsibility for the content presented.
NAU is co-financed by the Recovery and Resilience Plan (PRR).
