Workshop: AI and social justice in climate communication

Setting

The 10th Environmental Observation Conference took place in Bern on the 27th and 28th of November. The conference target group was mainly Environmental experts (environmental agencies and research institutions), communications experts (authorities, NGOs), media professionals, data officers, and AI experts, government actors (policymakers). Because the overall conference theme was artificial intelligence, the workshop focused on how marginalized groups can be supported and included in the transition away from combustion engines. It also examined what role artificial intelligence might play in this transition.

Workshop outline

The workshop consisted of three sessions and began with a brief presentation of the overall topic, followed by a discussion of the marginalized group’s scenario and context. The marginalized group was, in this context, defined as lower-income groups, as in previous measures by the German government; people with higher incomes mainly profited from the subsidies. For the workshop, it was not possible to directly include the marginalized groups. Still, the approach was to use existing knowledge and identify gaps in how environmental experts approach them. Afterwards, the mapping was explained and each of the questions, which were the following:

1. “Which AI tools could break down barriers to participation?”
2. ““Where does AI increase the risks of exclusion for certain groups?”
3. “Which analog methods work in practice? (Social associations, businesses, local initiatives)”
4. “What forms of participation would be needed?”

General outcomes

Overall, the outcomes of the three workshops showed that participants struggled to find ways to communicate and involve the group in discussions of climate change mitigation. Therefore, the workshop’s observations amplify the need for policymakers to include marginalized groups more in debates on social justice. However, despite their concern, the groups still came up with many ideas and had discussions within those 30 minutes. The ideas often aimed to make the electric vehicle and its benefits more tangible by showing how they would be used, for example, a scenario-based AI bot, where people could type in their daily driving patterns, or in the analog world by providing pop-up test drives in their neighborhoods. Further, the outcomes support the findings from the chapters “segmentation” and “norms and social identity” by using credible messengers and approaching communication within segments. As the majority of responses focused on everyday interactions in their everyday situations, ranging from the “Stammtisch” to cooperating with local sports associations. The reliance on everyday interaction to reach the group aligns with Herlo’s (2024) findings, which indicate that marginalized groups in social and economic settings may have difficulties with digital participation. Therefore, social participation should be seen as the leverage and the role of an AI tool rather than a supplement or assistant. The need for policymakers to consider ethical considerations in their communication was highlighted by concerns across all three workshop sessions that the current data the AI is using may widen existing inequalities due to imbalanced training data. The current imbalance is further supported by the author Gabriel (2026), who warns that biased AI training data may worsen it. This concern was strengthened by the rise, within two sessions, of the possibility of a lack of trust in AI. These two findings may correlate and be considered when using AI tools within the process.