Mr. Eoin O'Reilly:

That might have been me, so apologies for using that term. It means where the AI gets it wrong. We have discussed the potential of AI and human-like interactions, but if models are not trained appropriately and fed with the right data, they can hallucinate like humans do every now and again.