AI chatbots shouldn’t decide your vote
Ahead of the snap elections in the Netherlands, the Dutch Data Protection Authority warned against using AI chatbots for voting advice, citing bias and unreliability. A study found that chatbots tended to favor larger parties regardless of user input, while smaller parties were underrepresented.
OpenAI tackles bias in ChatGPT
OpenAI has released new results from an internal test aimed at reducing political bias in ChatGPT. According to the findings, the GPT-5 models are the most objective versions yet. The test evaluated how ChatGPT handles provocative questions across 100 politically sensitive topics.
AI models can reinforce misinformation and bias
New tests reveal some worrying trends in AI models. This became especially clear with Elon Musk’s “Grok,” which, after a political retuning, suddenly began spreading antisemitic and violent content.
AI models can reinforce misinformation and bias
New tests reveal some worrying trends in AI models. This became especially clear with Elon Musk’s “Grok,” which, after a political retuning, suddenly began spreading antisemitic and violent content.