Pearson Plc

In our view, the first Global Dialogue on AI Governance would be a success if it delivers concrete, shared foundations for governing the environmental impacts of AI, rather than high‑level principles alone. First, success would mean agreement on a core set of standardised metrics to measure and report the impacts of AI models. This includes consistent, decision‑useful metrics covering energy consumption, carbon emissions, water use and, where relevant, hardware and lifecycle impacts. Today, companies and users lack comparable data, which undermines accountability and slows meaningful action. A common measurement baseline, aligned with existing climate and sustainability reporting standards, would be a critical step forward. Second, the Dialogue should materially advance transparency and information‑sharing across the AI ecosystem, with a specific focus on infrastructure and cloud service providers. Progress on sustainable AI is constrained by limited disclosure from major hyperscalers. Encouraging, or mandating over time, greater transparency from providers such as Microsoft, Google Cloud Platform and AWS on AI‑specific energy, water, and emissions data would unlock better decision‑making by downstream users and customers, and enable more credible reporting and target‑setting. Third, success would mean converging on a standardised set of sustainability‑related targets that organisations involved in AI development and deployment should commit to. These targets should cover efficiency improvements, use of renewable electricity, reduction of water stress impacts, and continual optimisation of models and workloads. Shared expectations on targets would help shift the industry from voluntary, fragmented action towards a more level playing field and genuine progress at scale.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

These priorities reflect our focus on making AI scale responsibly from an environmental and societal perspective. As AI adoption accelerates, safety, trust and oversight are critical to avoid unintended consequences becoming locked into systems and infrastructure at scale. Transparency and accountability are especially urgent, as the lack of comparable, decision-useful data on AI energy, water and emissions impacts currently limits informed choices by organisations using AI. Capacity-building is equally important. Many organisations are deploying AI faster than they can understand or manage its broader impacts. Building shared capabilities around measurement, target-setting and governance is essential to ensure sustainability considerations are embedded in practice, not treated as an afterthought. Finally, the wider social, economic and ethical implications of AI matter because environmental impacts intersect with questions of equity, access to infrastructure and local constraints such as water stress or energy availability. Addressing these dimensions together is necessary to ensure AI supports sustainable development rather than exacerbating existing pressures or inequalities.

Yes. A major cross-cutting issue that is not explicitly captured is the environmental impact of AI across its full lifecycle, including energy use, carbon emissions, water consumption and hardware intensity. These impacts cut across all themes but are often treated as secondary or indirect, despite AI's rapidly growing resource footprint and its potential to conflict with climate and environmental objectives if left unmanaged. A related and emerging issue is the systemic over-production and storage of non-relevant data to train and operate AI systems. Large volumes of low-value or unused data are stored, processed and replicated, driving unnecessary energy and infrastructure demand. This "data waste" problem is rarely addressed directly, yet it significantly amplifies AI's environmental footprint and operational inefficiency.

As an education company operating across multiple regions, current governance gaps around AI are having tangible impacts on how we deploy AI responsibly at scale. One of the most significant challenges is the lack of clear, consistent expectations on transparency and environmental impact across jurisdictions. While expectations around safe, trustworthy and transparent AI are rising, organisations like Pearson still face limited visibility on the environmental footprint of AI infrastructure, particularly energy, water use and emissions linked to cloud‑based AI services. This makes it difficult to fully integrate AI into climate strategies and to align AI adoption with existing environmental targets, despite strong internal governance frameworks. Capacity gaps are another constraint. AI is being adopted rapidly across education, but many organisations lack the tools, data and skills to assess AI impacts holistically, spanning ethics, environmental performance, learner outcomes and equity. This increases governance complexity and the risk of fragmented or reactive approaches, especially in highly regulated education markets. At the same time, there are important opportunities. Advances in responsible AI governance, impact assessments and AI risk management frameworks provide a foundation to embed environmental and social considerations more systematically across product design and operations. Education companies are also well‑placed to contribute to AI capacity‑building by translating governance expectations into practical tools and learning for educators, institutions and learners.

The AI Dialogue can help by creating a shared space where countries and sectors can align on a small number of practical expectations, rather than continuing with fragmented and sometimes conflicting approaches to AI governance. Its main value is in supporting convergence: sharing lessons, surfacing areas of early agreement, and encouraging common approaches on issues like safety, transparency and sustainability. This is especially important for organisations operating internationally, where inconsistent rules and limited data make responsible AI harder to implement in practice. The Dialogue can also help rebalance international cooperation by ensuring that perspectives beyond large technology providers and a few regions are reflected, and by supporting capacity‑building where governance and technical capabilities are still developing. If it stays focused on enabling collaboration and practical outcomes, the Dialogue can complement existing initiatives and help turn broad principles into action that works across borders and sectors.

The AI Dialogue should build on existing work such as the UN Global Digital Compact, UNESCO's AI Ethics framework, OECD AI Principles and emerging standards and risk frameworks, rather than duplicate them. These initiatives already provide solid foundations on issues like trust, human oversight and responsible use, but they are developing in parallel and are not always well connected in practice. The added value of the AI Dialogue is its ability to bring these efforts together in one multilateral space, create visibility on how they relate to each other, and identify where alignment or convergence is realistically possible. For organisations operating globally, this coordination matters more than the creation of new principles. The Dialogue can also help connect governance discussions with implementation realities, including environmental sustainability, transparency of AI infrastructure, and capacity‑building for sectors that are adopting AI rather than developing it. By linking high‑level frameworks with concrete challenges faced by users of AI, such as education, public services or SMEs, it can help ensure that global governance remains grounded and relevant. If it focuses on coordination, shared learning and practical follow‑through, the Dialogue can strengthen the overall AI governance ecosystem without adding complexity.

The Dialogue should be designed so engagement does not depend on being physically present. This means enabling ongoing written inputs, remote participation, and clear feedback loops beyond the event itself. Stakeholders should be able to contribute through online consultations, thematic working groups, and follow‑up activities that continue between Dialogue sessions. Publishing clear summaries, priorities and next steps after each Dialogue is essential to allow organisations to engage, react and contribute over time. A light but structured mechanism for ongoing interaction would ensure the Dialogue remains inclusive, accessible and relevant to a wider range of actors, including those unable to attend in person.

N/A

Meaningful engagement would be better supported by formats that go beyond traditional panels. Smaller, facilitated discussions focused on concrete questions or case studies can encourage more honest exchange and learning across stakeholder groups. Hybrid formats are also essential. Live‑streamed sessions with the ability to submit inputs, join moderated online discussions, or contribute asynchronously would allow stakeholders to engage even if they cannot attend in person, reinforcing the Dialogue's inclusive mandate. Finally, light thematic working groups or online forums between Dialogue sessions could help maintain momentum, allowing ideas raised during the Dialogue to be tested, refined and built on over time rather than remaining one‑off contributions.

One useful example is the adoption of management-system approaches to AI governance, such as ISO/IEC 42001, the first international standard for Artificial Intelligence Management Systems. ISO 42001 provides a structured, auditable framework for governing AI across its lifecycle, covering accountability, risk management, transparency and continuous improvement, and is applicable to organisations that develop, deploy or use AI systems. Other effective practices include responsible AI frameworks that embed impact and risk assessments, human oversight and data governance into day-to-day decision-making, as well as cross-functional governance structures that connect legal, technical, ethical and sustainability considerations. Platforms and approaches that support measurement and reporting, particularly around AI performance, risk and environmental impacts, are also critical enablers, as they translate high-level principles into operational decisions.