A recent study by computer and information science researchers from the UK and Brazil has raised a red flag about ChatGPT’s objectivity.
They found that ChatGPT tends to lean towards the left side of the political spectrum, indicating a significant political bias.
This study, published in the journal Public Choice, discusses how this bias in AI-generated content can perpetuate existing biases in traditional media.
It’s not just a theoretical concern; it could affect policymakers, media outlets, political groups, and educational institutions.
The researchers used questionnaires to gauge ChatGPT’s political orientation and found it consistently favoring Democratic-leaning responses, even when impersonating both Democrats and Republicans.
Surprisingly, this bias extends beyond the US and affects responses related to Brazilian and British politics.
While pinpointing the exact source of this bias remains a challenge, the researchers suspect both the training data and the algorithm itself play a role.
OpenAI has yet to respond to these findings, but this study adds to a growing list of concerns regarding AI, including privacy and identity verification issues.
As AI-driven tools like ChatGPT become increasingly influential, it’s crucial to remain vigilant and critically assess their content.
This study reminds us of the importance of developing and deploying AI technologies in a fair and balanced manner, free from undue political influence.