Generative AI can speed up sustainable product development, but it also risks creating new problems: it uses lots of energy, can invade privacy, the data can be biased, it may benefit only a few, and may push us to “do the wrong thing faster.” real sustainability requires more than just optimization, it needs creativity, fairness, and long-term responsibility.
1. Data Privacy Risks
Circular economy (CE) systems often require the collection and sharing of large amounts of data across multiple stakeholders to “close the loop.” This includes tracking products via IoT devices or detailed usage data.
Such data collection poses privacy risks, especially when it involves personal or geospatial data. Consumers may be uncomfortable with how their data is used, potentially undermining trust in CE initiatives.
Data analysis by AI can infer sensitive information about behaviours or habits (e.g., smart meters revealing household routines), creating further privacy and surveillance concerns.
Risks multiply when multiple organizations share data, increasing the chance for data breaches or misuse, especially in highly interoperable ecosystems.
2. Algorithmic Bias
AI-driven business models in the CE, such as dynamic pricing and automated matching, can propagate unfair or discriminatory outcomes. For instance, dynamic pricing algorithms might offer different prices based on proxies for protected characteristics (like ZIP codes, correlating to ethnicity or income).
Biases may be embedded within AI models either directly (through data) or indirectly (via proxies), leading to systemic discrimination at scale. Resolving these issues is difficult due to the opacity (“black-box” nature) and complexity of AI systems.
3. Economic Inequality and Exclusion
At an international level, the adoption of AI in the circular economy could widen the digital divide. High-income (Global North) nations are more likely to benefit from advanced AI deployment for CE, while Global South countries remain relegated to roles in raw material extraction or low-skill tasks.
Domestically, shifts toward CE and the use of AI can lead to polarization in job markets (e.g., more high-skill design/maintenance roles, fewer low-skill manufacturing roles), increasing wage and geographic inequality.
Digital platforms and AI-based services may exclude non-digital populations (like the elderly or less informed), furthering social exclusion.
4. Epistemological and Environmental Risks
There is a risk of oversimplifying complex ecological systems when deploying AI, potentially optimizing for narrow environmental goals while ignoring important trade-offs (such as the environmental toll of digital infrastructure or new technologies themselves).
If AI is deployed without a full scientific understanding, it might inadvertently cause new harms (e.g., recycling processes that create unexpected waste or pollution).
Significant energy consumption by AI and digital infrastructure (like data centers) can undermine environmental goals if not properly managed.
While the environmental cost of AI is real, its “climate potential is far greater if directed wisely.” – The Economist 2025
AI tasks are energy-intensive and data centers require significant electricity, leading to notable emissions increases among tech giants.
Despite a single AI query consuming far more energy than a traditional web search, overall AI energy use remains a small share of global consumption.
AI is particularly promising in “hard-to-clean-up” industries like heavy manufacturing, shipping, and agriculture where conventional climate strategies have struggled.
In manufacturing, AI enables more efficient energy use and predictive maintenance.
In agriculture, AI supports precision farming, reducing fertilizer and water waste.
AI also accelerates discovery of new battery materials and improves carbon capture technologies.
AI systems enhance climate modelling, grid management, and emission detection, making it possible to optimize complex systems for maximum carbon reduction.
Its biggest impact may be in helping industries and governments achieve far faster progress towards decarbonization goals, if AI infrastructure itself is powered by clean energy.
Policymakers and investors need to guide the direction of AI towards climate-positive uses, e.g., through incentives for clean data centers and transparent AI emissions reporting.
The article urges integrating renewable energy into data centers and prioritizing AI deployment in sectors with high emissions
Generative AI tools can shorten physical product design life cycles significantly and spark innovation, but the knowledge and discretion of design experts are necessary to mitigate potential pitfalls.
AI as a Companion in World-Making:
AI should not just be a tool to make predictions or optimize efficiency. Instead, it should act as a companion or partner to humans and non-humans, participating in the ongoing, collective process of imagining and enacting sustainable futures.
Embrace Relational Practices:
AI’s contributions should support dialogue, empathy, narrative, and attunement to context rather than mere data extraction and operational dashboards. AI can help give “voice” to non-human actors (e.g., using generative models to simulate the perspectives of wildlife), expand ongoing conversations, and enable learning from multiple, plural futures.
Foster Imagination and Hope:
Rather than eliminating uncertainty, AI can create spaces for imagination, exploring new possibilities, “what if” scenarios, and alternative futures. This involves collaborative storytelling, interactive simulations, and creative engagements where uncertainty is seen as potential rather than a risk to manage.
Response-Ability for Humans and More-Than-Humans:
AI systems should help facilitate ethical responsiveness, not just control environments, but cultivate caring, attentive relationships among all actors in the system, considering interdependence across time, space, and species.
Recognize and Address Material and Ecological Costs:
A responsible, relational approach to AI acknowledges the material and ecological footprint of digital technology (energy demands, resource extraction, labor) and seeks to minimize harm. It insists on remaining “present” with these tensions, seeking lighter footprints and reparative practices.
Enable Participatory Futures:
AI for sustainability is best deployed in participatory, community-centered ways, such as co-designing with local communities, supporting collective explorations of possible futures, and democratizing access and narrative power around what “sustainable” looks like.