AI models struggle to identify nonsense
The AI models that power chatbots and other applications still have difficulty distinguishing between nonsense and natural language, according to a study released on Thursday.
The researchers at Columbia University said their work revealed limitations of AI models and suggested it was too early to let them loose in legal or medical settings.
They put nine AI models through their paces, firing hundreds of pairs of sentences at them and asking which were likely to be heard in everyday speech.
The research, published in Nature Machine Intelligence journal, then weighed the AI answers against the human answers and found dramatic differences.