The ChatGPT Artificial Intelligence (AI) tool, trained to write a dissertation in philosophy, produced an academically acceptable version of a baccalaureate subject in France, according to an experiment conducted in Paris this Wednesday.
“Is happiness a matter of reasoning?” was the topic analyzed by this ‘software’, which ‘faced’ the philosopher Raphaël Enthoven in a ‘meeting’ organized by a business and technology school.
Result: ChatGPT scored 11 out of 20 and Raphaël Enthoven 20/20, the agency France-Presse (AFP) reported.
These numbers were awarded jointly by philosophy teacher and writer Éliette Abécassis, and a secondary school teacher known by the pseudonym “Serial Thinker” on the social network TikTok.
If the works had been judged in a ‘blind test’, the jury would have guessed “at the first words” who wrote what, Abecassis stressed.
“In the ChatGPT version there is not even a problem (…) Often they are very long sentences, quite hollow at the end, where there is no content, where we do not understand the arguments (…) quotes to try to shine”, he underlined Lev Fraenckel, known as “Serial Thinker”.
“This is not a philosophy at all, it is not stringing together fancy sentences,” he added, pointing out that “the reference to the authors is very weak, because there are mistakes in it.”
The Paris School of Technology and Business, which organized the “duel,” refined a very long question to ask the “software” for the classic forms of the thesis, and also suggested authors refer to the ChatGPT.
ChatGPT quoted Aristotle, Kant, Freud, Nietzsche and Camus, with vague and brief references.
The ‘software’ showed some stylistic boldness, but, trained not to have an opinion, it did not interfere with the issue.
The conclusion states: “There is no universal answer, but a myriad of paths to happiness (…) Happiness may well be a matter of reason… and much more”.
Raphaël Enthoven believes that philosophers are among the professionals least likely to be replaced by artificial intelligence.
“The teaching of philosophy is not threatened (…) I bet we could have steered [ao ChatGPT] a whole book of recommendations, without making him a philosopher,” he underlined.
The work elaborated by man, performed in an hour and a half, ends with “the urgency and interest in thinking about reason itself and its activity as happiness”.
AI systems are technically very complex and fascinate in the same way that they worry.
The general public discovered its enormous potential late last year with the launch of the editorial content producer ChatGPT, from the Californian company OpenAI, which can write essays, poems or translations in seconds.
Source: DN
