The AI chatbots ay kayang makaapekto sa political views ng mga tao, ayon sa bagong studies. Ipinakita ng research na kahit maikling usapan lang sa isang partisan chatbot ay puwedeng magbago ng pananaw ng botante.
In experiments using generative AI such as GPT-4o and DeepSeek, it has been found that it can push some Trump supporters towards Kamala Harris by almost four points on a 100-point scale before the 2024 US elections. In Canada and Poland, the impact was even greater—up to 10 points.
According to Professor David Rand of Cornell University, this change is enough to affect voting decisions. He said that “one in 10” in Canada and Poland changed their voter choice, while “one in 25” in the US.
The study also showed that the most effective way for chatbots to persuade was to be polite and provide evidence. Notably, the impact was weaker for bots that were not allowed to use facts. But not all AI “evidence” is correct—some of it is misinformation, especially when it favors right-leaning candidates.
It was also discovered that even after a month, half of the effect remained in Britain and one-third in the US. For Rand, it is rare in social science to have an effect that lasts this long.
