Sanoma Learning’s approach to sustainability in AI covers the full scope of environmental, social and governance-related (ESG) aspects. We aim to leverage AI’s potential to enhance learning, while ensuring ethical and transparent use and mitigating AI’s adverse environmental impacts.
AI can create educational value in multiple ways. In digital learning products, AI helps teachers respond to different learner needs more effectively. It enables personalised learning and makes differentiation easier in everyday classroom work. It can also help free teachers’ time for feedback, pedagogy, and student support.
For creating learning materials, AI can help create, adapt and localise content faster, which makes it easier to scale diverse and accessible learning materials across subjects, languages and teaching contexts. Using AI to increase productivity in software development and general internal processes lets our people focus more on what matters: creating high-quality learning products together with teachers.
When using AI-tools or developing AI-enabled products, Sanoma Learning applies its ethical AI principles. These principles also form the basis for our Privacy, Security and AI by Design process, which ensures compliance with regulatory requirements such as the EU AI Act across our company. Specifically, the process comprises privacy assessments for all AI use cases, an AI initiatives inventory, and structured analysis and management of the impact of AI towards students, teachers and other stakeholders.
Furthermore, Sanoma Learning assesses the environmental impacts of its AI use throughout the lifecycle of products and services, focusing on the parts we can influence. The environmental footprint of AI primarily depends on the electricity mix used in the data centres where models are trained and run, as well as on overall resource efficiency, particularly with regard to water use. For this reason, we select trusted technology partners that offer low-emission and resource-efficient solutions.
Whenever possible, we keep data storage and processing within the EU, where the share of fossil-free energy is higher than in many other regions. As we do not train or host AI models ourselves, our primary environmental impact stems from model inference, meaning the resources used for generating outputs. While the energy consumption for creating a single output is small, the cumulative environmental impact of inference can become significant when AI is used at large scale.
This is why environmental considerations are a key factor in our approach to AI. In other words, we aim to look beyond what AI can do for learning and consider the resources it consumes. This includes energy‑efficient technical design, selecting partners with transparent energy and emissions reporting, using smaller models for simpler tasks, and avoiding AI use when benefits are marginal.
