AI hallucinations exposed in analysis of Estonian poet Heiti Talvik's works

A reader raised concerns that AI-generated German source texts cited in an analysis of Estonian poet Heiti Talvik's poems may have been fabricated hallucinations. Journalist Aivar Kull addresses the issue in a correction to his earlier history column published on May 26.

A sharp-eyed reader contacted the editorial team after reading Aivar Kull's history column published on May 26, raising a serious question: could the alleged German-language source texts that an artificial intelligence tool produced in connection with the poetry of Estonian poet Heiti Talvik simply be convincing but entirely fictional AI inventions?

The concern points to a well-documented phenomenon in the field of artificial intelligence known as "hallucination" — where large language models generate plausible-sounding but factually incorrect or entirely fabricated information. In literary and historical research, such errors can be particularly damaging, as false source attributions may go unnoticed and spread through subsequent references.

Aivar Kull acknowledged the issue in a follow-up correction to his original column, signalling that the question deserved serious attention. The case serves as a cautionary example of the risks involved when AI tools are used in humanities research without rigorous verification of the sources they produce.

Heiti Talvik (1903–1945) is considered one of the most significant Estonian poets of the interwar period. Questions about the origins and possible foreign-language influences on his work are of genuine scholarly interest, which makes the accuracy of any claimed source texts all the more critical.