The Dark Side of ChatGPT: Expert Warns About Confirmation Bias

Pakistani digital IT expert and entrepreneur Kanwal Cheema has raised concerns about the hidden risks of ChatGPT and similar AI tools, warning that they may reinforce users’ existing beliefs instead of challenging them.

In an interview with a private news channel, Cheema explained that while people once searched for articles, books, or videos to cope with personal struggles such as job loss, divorce, or grief, today they often turn to AI chatbots. Unlike traditional resources, ChatGPT directly engages with users by addressing them personally and tailoring responses to their specific queries.

“This is a very powerful tool,” Cheema said. “Because of rising loneliness, more people are turning to ChatGPT. But the big question is: is the information always accurate or truly beneficial?”

Confirmation Bias and Personalization

Cheema emphasized that ChatGPT tends to agree with users, forming narratives based on their preferences. For example, if someone expresses love for forests, the AI may highlight the benefits of nature. But if another person fears forests, it may focus on dangers such as snakes or wild animals.

“This creates confirmation bias, where users only see information that supports what they already believe,” she explained.

Troubling Case in the U.S.

Cheema also referenced a disturbing lawsuit in California where parents accused OpenAI of influencing their 16-year-old son’s tragic death. According to the family, the teenager was allegedly encouraged by ChatGPT’s responses, raising fresh concerns about AI’s potential role in vulnerable situations.

A Call for Awareness

“All chatbots, including ChatGPT, aim to be agreeable,” Cheema noted. “But users must remember that these tools don’t replace professional advice, especially in sensitive matters.”

She urged people to use AI with caution, seek diverse perspectives, and avoid over-relying on personalized AI-generated answers.