tech
Naučnici upozoravaju na pogrešne informacije
U Belgiji je nedavno pokrenuta debata oko upotrebe veštačke inteligencije, jer je Petra de Suter (Sutter), rektorka Univerziteta u Gentu, koristila veštačku inteligenciju da napiše svoj uvodni govor za akademsku godinu prošlog septembra, ali njen tekst sadrži greške i dva izmišljena citata – to je ono što je poznato kao „halucinacija“, preneo je belgijski javni servis RTBF.
TL;DR
- Petra de Suter, rector of Ghent University, used AI to write her opening speech for the academic year.
- The AI-generated text contained errors and two fabricated quotes, a phenomenon called 'hallucination'.
- One fabricated quote was attributed to Albert Einstein, stating 'Dogma is the enemy of progress,' which he never said.
- Professor Hugues Bersini explained that AI statistically combines words, creating plausible but false statements.
- AI hallucinations are less common in areas with limited vocabulary, such as programming and mathematics, but more frequent in diverse text fields like essays and speeches.
- Users must critically evaluate AI-generated information and verify its accuracy, as AI does not understand truth or check its sources.