AI Chatbots Are Biased Against Women and Climate Justice, New Study Reveals
Groundbreaking research exposes how ChatGPT and other AI tools are reshaping climate conversations—and leaving women and marginalized communities behind
By SheSight Editorial
As millions turn to AI chatbots like ChatGPT and Claude for information about climate change and environmental challenges, a groundbreaking study from the University of British Columbia reveals these tools harbor significant biases that disproportionately affect women and marginalized communities—while actively avoiding solutions that challenge existing power structures.
The Hidden Bias in Your AI Assistant
The research, published in Environmental Research Letters, analyzed 1,512 chatbot responses across multiple AI platforms and nine environmental challenges. What researchers found should concern anyone who cares about climate justice: AI chatbots are not neutral information sources. Instead, they’re reinforcing systemic biases that could actually slow progress on the climate crisis.
“Chatbots provide concise and relevant responses within a single textbox, often in an authoritative tone that can imbue them with an air of wisdom,” explains lead researcher Dr. Hamish van der Ven. “The danger lies in the way chatbots mask bias.”
Women Erased from Climate Conversations
Perhaps most striking for SheSight readers: women are mentioned only 36 times across all responses about who is vulnerable to environmental challenges—compared to indigenous groups mentioned hundreds of times. Black communities fared even worse, with just 15 mentions.
This isn’t just about representation. It reflects a deeper problem: AI chatbots are systematically downplaying how climate change disproportionately impacts women, who comprise 80% of people displaced by climate disasters according to the UN.
The bias extends to knowledge production itself. The study found:
- 96% of sources cited had male first authors
- 100% of citations came from Global North-based researchers
- Zero references to indigenous or local knowledge systems
- A heavy preference for quantitative over qualitative research
“These platforms are not value-neutral,” notes van der Ven. “They offer a particular interpretation that may diverge from expert understandings of environmental challenges.”
Big Finance Gets a Free Pass
While AI chatbots readily identify governments as responsible for environmental problems—mentioning them 236 times—they mentioned investors, finance, and capital just 66 times. Businesses and corporations appeared 168 times.
This matters because fossil fuel companies, banks, and investment firms are major drivers of the climate crisis. Yet AI is systematically underplaying their role while putting the onus on government regulation and individual action.
The Incrementalism Trap
Perhaps most concerning: when asked about solutions to environmental challenges, chatbots consistently avoided suggesting radical or transformative changes.
What AI won’t tell you:
- Zero mentions of challenging capitalism or economic growth
- No references to degrowth or alternative economies
- No suggestions of civil disobedience or protest
- Barely any mentions of decolonization
- Almost no use of the term “environmental justice”
Instead, chatbots recommend incremental approaches like public education campaigns and sustainable technology deployment—the same strategies that have failed to prevent us from hurtling toward catastrophic warming.
With scientists warning that global temperatures will cross the critical 1.5°C threshold between 2033 and 2035, this backward-looking bias could be catastrophic.
Why Environmental Justice Disappeared
The study found that chatbots overwhelmingly avoid connecting environmental challenges to broader social justice issues. Terms like “justice,” “equity,” “racism,” “sexism,” “discrimination,” and “colonialism” were conspicuously absent from most responses.
When chatbots did acknowledge adverse impacts of environmental changes, they mentioned consequences for indigenous communities far more than for women or people of color—despite extensive research documenting how these groups face compounding climate vulnerabilities.
“The clever bundling of multiple opaque sources into a single synthetic response makes it difficult to identify novel forms of bias,” the researchers explain, “but also allows chatbots to amplify and normalize pre-existing societal biases.”
What This Means for Climate Action
As policymakers, educators, and the public increasingly rely on AI for information about climate change, these biases aren’t just technical glitches—they’re actively shaping how we understand and respond to the crisis.
The research team warns that AI’s tendency toward “dangerous incrementalism” could actually impede climate action by:
- Circumscribing the realm of possible solutions
- Delaying necessary radical changes to economic and political systems
- Reinforcing ineffective status-quo approaches
- Obscuring the connections between climate and social justice
The Tech Bros’ Climate Blind Spot
The findings challenge claims from tech leaders like Elon Musk that AI chatbots are too “woke.” On the contrary, the research reveals right-leaning biases when it comes to environmental issues—avoiding calls for systemic change, separating environmental challenges from social justice concerns, and placing responsibility overwhelmingly on governments rather than corporations or investors.
What We Can Do
The researchers call for action from multiple stakeholders:
For users: Develop digital literacy. Question AI outputs critically, especially regarding solutions and forward-thinking. Don’t trust the authoritative tone.
For educators: Teach students to recognize AI bias and seek diverse sources, particularly from women, Global South scholars, and marginalized communities.
For tech companies: Increase transparency about training data composition and provide traceability for chatbot outputs.
For regulators: Weigh AI’s positive impacts against adverse effects, including bias propagation.
The Bottom Line
AI can be a powerful tool for addressing environmental challenges—but only if we understand and actively counter its built-in biases. As it stands, these tools are reinforcing exactly the kind of thinking that got us into the climate crisis: incremental, top-down, justice-blind approaches that protect existing power structures.
For women and marginalized communities already bearing the brunt of climate change, AI’s biases represent yet another barrier to climate justice. And in a world racing against time to prevent catastrophic warming, we can’t afford to let artificial intelligence slow down the transformative change we desperately need.
The study, “Does artificial intelligence bias perceptions of environmental challenges?” was published in Environmental Research Letters (2025) by researchers from the University of British Columbia.
Read the full paper: https://doi.org/10.1088/1748-9326/ad95a2
About SheSight
SheSight is a global digital media platform amplifying the voices, leadership, and stories of women around the world. We publish thoughtful features, expert commentary, and real-world insights on gender, leadership, equity, and impact.
Support Our Mission
Love what we do? Help us continue telling stories that matter.
Support us on Ko-fi every contribution helps us grow this movement.
If you are here, reading this article, you are probably interested in elevating your aspirations, taking bold leadership and leading a more meaningful life. At ShePower Club, we offer worldwide sisterhood for driven women like you! – A platform to connect, develop and amplify your voice
