ChatGPT and other AI bots made huge errors before Scottish election, study finds
Summary
A study by the thinktank Demos found that popular AI chatbots, including ChatGPT and Google Gemini, gave incorrect or misleading answers to many questions related to the recent Scottish election. The UK's Electoral Commission is calling for new laws to regulate AI platforms to prevent the spread of false information during elections.Key Facts
- Demos tested five free AI chatbots by asking 75 election-related questions about Scottish constituencies.
- AI tools gave wrong answers to 34% of all questions asked.
- ChatGPT was wrong in 46% of its answers, including inventing fake scandals and giving incorrect election dates.
- Replika made mistakes in 56% of answers, inventing candidates and scandals.
- Google Gemini made errors in 22% of answers, such as wrongly stating political positions and ongoing investigations.
- Grok, an AI linked to Elon Musk’s platform X, had the lowest error rate at 9%, but its external links were often poor or irrelevant.
- 20% of British voters used AI tools for election information, which could be about 10 million people.
- The Electoral Commission’s chief executive wants laws to hold AI companies accountable and give regulators more power during elections.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.