In a media study, researchers found that liberal-leaning media is more oppossed to artificial intelligence than conservative-leaning media.
Artificial intelligence is slowly making its way into our daily lives, but it continues to be a divisive technology. New research has discovered that articles from liberal-leaning media in the United States have more of a negative attitude towards AI than articles from conservative media.
A study published in the journal Social Psychological and Personality Science found this difference in coverage and also found a reason for this. The research focused on American publications so the results may not be globally applicable.
Liberal-leaning media’s opposition to AI can be attributed to them being more concerned with magnifying social biases in society like race, gender and income disparities than conservative media. This research was focused on American publications.
“Our research found that the liberal media’s aversion to AI can be attributed to the concerns about AI magnifying the social biases in society. In addition to social bias concerns, we also tested for privacy concerns and employment concerns as other reasons behind the difference in sentiment toward AI between liberal and conservative media. First, it is possible that liberal media are more averse to AI because they are more concerned about invasions of privacy. Alternatively, it is also possible that liberal media are more averse to AI because conservative media values the efficiency and corporate productivity caused by AI,” said Angela Yi, co-author of the paper, to indianexpress.com in an email interaction.
But after running the analysis, they found that privacy concerns and employment concerns do not explain why liberal media is more averse to AI than conservative media. Essentially, those alternative explanations do not really explain the aversion.
These findings could have important implications for the future political discussions around AI. This is because media sentiment is often used as an indicator of public sentiment. That in turn, can impact policymakers’ stances.
How the media was studied
To explore media sentiment towards AI, researchers put together a list of articles about AI from many media outlets. They then measured the partisan sentiment of each outlet using the ratings found on the Media Bias Rating Chart from AllSides. Later, they used articles from a mix of liberal-leaning outlets like The New York Times and The Washington Post along with more conservative-leaning outlets like The Wall Street Journal and New York Post.
The articles they downloaded were selected from the outlets based on specific criteria, like the usage of key terms like “algorithm” or “artificial intelligence,” from May 2019 to May 2021. Do note that this was a long time before the launch of ChatGPT in November 2022.
After creating a dataset of over 7,500 articles, they performed an emotional tone analysis on each story using an automated text analysis tool. The tool calculated the difference between the percentage of positive emotion words and the percentage of negative emotion words in a text. This difference was standardised on a scale of 0 to 100 to make the emotional tone measure.
What the researchers found
The study found that articles from liberal-leaning media have more of a negative sentiment towards AI than articles from conservative media sources. In essence, liberal media seemed to be more opposed to AI than conservative-leaning media.
This opposition from liberal-leaning outlets can for the most part be attributed to concerns with AI magnifying social biases in society, including race, gender, and income disparities. This was not as much of a concern for conservative media. Interestingly, the researchers also examined whether media sentiment towards AI changed after George Floyd’s death.
“The media sentiment towards AI became more negative after George Floyd’s death. Since Floyd’s death sparked a national conversation about the social biases in society, we assumed that this event may have also heightened social bias concerns in the media. As expected, we find that his death heightened social bias concerns in the media articles about AI. Hence, this increase in social bias concerns in the media articles about AI led the sentiment in these articles to be more negative,” said Yi.
The co-authors of the paper stress that this research is “descriptive rather than prescriptive,” and that it does not take a stance as to the right way of discussing AI. Essentially, they are not saying that either the liberal or the conservative media is doing it right. Instead, the researchers just showed how there is a clear difference in the media sentiment.