Most of today’s social media users are at least aware of the almighty algorithm on their platforms, known for serving up content geared toward their interests. But what about on search engines like Google?
When researchers from Columbia Business School searched for climate change images in different countries, they discovered just how much the results are tailored to a region’s prevailing viewpoints. On major societal issues like the climate crisis, that influence can have a profound impact.
Key Takeaways:
- Google Images search reflects and amplifies preexisting national sentiments on climate change. In countries whose citizens are more concerned by climate change, the image results tend to be more emotionally arousing compared with the results in countries with lower levels of concern.
- Exposing viewers to emotionally charged images from countries with high climate concern increases their climate concern, support for climate policy, and intentions to take pro-environmental action.
- The findings have implications for policymakers, who might consider leveraging similar platforms or imagery to raise public climate consciousness and support for climate action. At the same time, there’s a risk of misinformation or emotional overload, so climate communication strategies must be carefully designed.
When you enter climate change into a Google search and select Images, what results do you get? A photo of a parched, cracked landscape bordered by smokestacks spewing plumes into a hazy sky? Maybe an illustration of Planet Earth engulfed in flames? Or do you see less provocative images, like charts and graphs, a picture of a person in business attire, or a simple photograph of an iceberg? According to a new study from Columbia Business School, what you see all depends on where you live — and how your neighbors feel about climate change.
Effective mitigation of climate change depends on worldwide awareness and action. As such, recent research has focused on ways to increase people’s belief in human-caused climate change — a phenomenon with broad scientific consensus — and thereby boost sustainable behaviors and support for climate policies. Building on this type of research, the CBS study assesses the ways people’s conceptions of climate change are shaped. Specifically, it examines how the algorithm of Google Images search both reflects and influences public perception of the issue.
The Power of Algorithms
Anyone who uses social media has experienced algorithms that respond to their activity to shape the content the platforms show them. “It’s well known now that sites like X or Facebook are learning about our interests, our identities, the people that we’re connected to,” says Michael Berkebile-Weinberg, a teaching fellow in CBS’s Business Management Division. “In real time, we can see our feeds being updated by these things.”
People know less about the algorithmic elements of platforms like Google, however. “We have a different expectation for how search engines work,” Berkebile-Weinberg says. “Google is built to be asked questions and to give an accurate answer, so I think we have an expectation that they are slightly more objective.”
The Impact of Climate Concern
For this study, the researchers examined whether Google Images search results related to climate change were consistent across different users. “Are search engines giving us an objective view on critical social or news issues, or are they being shaped by what they think we want to know or what the people around us are interested in?” Berkebile-Weinberg says.
The researchers explored Google Images results across 49 different countries — some more concerned with climate change and some less so, according to a Gallup World Poll. They used VPNs to make it appear that the requests were originating from a particular country and a private browser to remove specific users’ preferences before entering search terms like climate change, climate crisis, and global warming. Then, they had a group of participants rate the emotionality of the images.
The results were straightforward, according to Berkebile-Weinberg: When a user searches for climate change imagery in countries that “care more” about climate change, Google yields emotional, evocative results, like a flaming planet or a split image of a tree, one half lush and green and the other half parched and dead. On the other hand, in countries whose residents told Gallup they were less concerned about climate change, the images Google returned from the same searches appeared less emotional — think pictures of graphs and icebergs, or maybe a few cartoon illustrations. Study participants rated these images as less emotional than the others. “It turns out, it doesn’t evoke many emotions for people when they see things like icebergs that are ostensibly smaller than they were a few years ago,” Berkebile-Weinberg says.
The researchers found consistent search results, whether a country’s leadership had taken steps to mitigate climate change or the area had recently experienced impacts of climate change. This supports the idea that image search outputs are driven by a country’s prevailing climate-change sentiment rather than its objective climate realities.
Influencing Action Through Emotion
Taking the experiment further, the researchers showed another set of participants images of climate change that came from searches conducted in countries high or low in climate concern and studied their reactions. They found that the people exposed to the more evocative images from countries with high preexisting concern became more concerned about climate change, more supportive of climate policy, and more likely to act pro-environmentally than those who saw the low-concern images that had been ranked as less emotional. This correlation persisted even when researchers controlled for participants’ political ideologies, suggesting internet search algorithms could influence sentiments on climate change, not just reflect them.
The findings have major implications for policymakers and organizations striving to mobilize citizens in the fight against climate change, suggesting the power of the algorithm could theoretically be harnessed to boost climate change awareness and policy support — both of which are critical to addressing the worsening climate crisis. At the same time, given the internet’s vast power to manipulate users through their emotions, as well as the risk of misinformation, Berkebile-Weinberg says the findings should offer a note of caution to web users. “Have a critical eye with what you are consuming,” he says, although he acknowledges that’s easier said than done. “That’s extremely hard when there are structures in our environments that are shaped by algorithms that we don't see.”
The research may also inform companies on how they should handle their role in shaping public opinion on major societal issues like climate change. “For a company such as Google, which has research teams in all sorts of different domains looking at their impact and how to make their product better, I think this work provides a good entry point for them to take stock of what product they should actually be designing — and, at the very least, how it should be advertised — because that would inform users,” Berkebile-Weinberg says.
Adapted from “Internet image search outputs propagate climate change sentiment and impact policy support,” by Michael Berkebile-Weinberg of Columbia Business School and New York University, Runji Gao of New York University, Rachel Tang of New York University, and Madalina Vlasceanu of New York University and Stanford University.
