AI hallucinations pose growing risks to professional accuracy

The rise of generative artificial intelligence has introduced significant risks regarding 'AI hallucinations'—instances where models produce false or erroneous information that appears highly convincing. Because these models prioritize generating likely sequences of text over factual verification, they can produce polished, confident-sounding answers that lack underlying accuracy.

In professional sectors, particularly law, these errors carry high stakes. In the United Kingdom, the Solicitors Regulation Authority has received dozens of reports regarding the misuse of AI, including the citation of non-existent legal precedents. In Lithuania, experts warn of 'vibe lawyering,' where non-professionals use AI to draft legal claims or contracts, potentially leading to invalid documents that fail to protect interests.

The real-world impact extends to government and consultancy work. For example, in Australia, a government report produced by Deloitte was found to contain incorrect and non-existent references after being flagged by academic researchers. Experts emphasize that accuracy is not just a property of the machine, but a requirement of the human process, necessitating rigorous judgment to verify AI-generated content.