ChatGPT, Bard or Grok: AI-controlled chatbots can distort facts, because they are trained to always use the word that is statistically most likely to follow in a sentence. However, they do not yet understand context.