Search
Close this search box.

Analyzing bias in content design resulting from human-AI collaboration

This image was generated by Midjourney AI

The study “Human favoritism, not AI aversion: People’s perceptions (and bias) toward Generative AI, human experts, and human-GAI collaboration in persuasive content generation” examines people’s perceptions and biases towards human-AI collaboration in content creation and investigates the presence of AI aversion.

To investigate this phenomenon the researchers have partnered with a reputable consulting firm to create advertising and persuasive content. Content was created by both AI (ChatGPT4) and professional content creators to compare the quality of design and, subsequently, bias generated by the content produced. 

“One might wonder whether a higher perceived quality of human-generated content in the ‘informed’ condition is due to a quality prime (i.e., knowing that the human creators are top industry professionals affects the ‘evaluation experience’) or due to ‘biased favoritism’ toward human experts.”

The study found that people tend to prefer content created by human experts, which results in a perceived increase in the quality of the content. However, it also showed that this preference is not due to a bias against AI-generated content, but rather human favoritism. The study did not find strong evidence of algorithm aversion, and it demonstrated that knowing that a piece of content is generated with the involvement of AI does not lower the level of satisfaction or willingness to pay for the content.

Share the Post:

Related Posts