Y Combinator co-founder Paul Graham has come under fire, mostly from Nigerians, following a ChatGPT text attribution theory he recently made on X (formerly Twitter).
Paul Graham Founder ChatGPT Text Theory
Graham posted about a cold email proposal he received from someone for a novel project. He immediately concluded the pitch was written by AI text generators like ChatGPT because it contained the word “delve,” which he claimed is rarely used in spoken English.
“No one uses it in spoken English,” Graham posted. “It’s one of those words like “burgeoning” that people only use when they’re writing and want to sound clever.”
Graham also shared a chart that depicted a correlation between the use of “delve” in articles and the inception of AI generators. The chart showed that the number of papers containing “delve” in the title or abstract has exponentially increased since 2022, the same year ChatGPT was launched.
Nigerians Slam Graham Over ChatGPT Theory
Thousands of X users, especially Nigerians, were quick to condemn the ChatGPT theory by Graham, saying that words like “delve” are not uncommon in the country and other nations with a British colonial past.
Some users also asserted that the rare use of such words is an American problem.
“The unfortunate thing is that the American vocabulary is primarily based on colloquialisms (what you call slang) and American contemporary idioms, to the extent that the sophisticated use of English diction and register is presumed to be AI-generated by people primarily familiar with the American worker,” an X user posted.
Spotting AI Writing is Getting Trickier
As time progresses, it becomes trickier to identify articles written by AI, given the unending batches of updates passed to the models. For some people, words like delve, tapestry, etc., have been a pointer because they are often overused by ChatGPT and its likes.
The words are not the problem. However, their repeated appearance in works produced by AI generators has led to theories like Graham’s that articles with such words are most likely written by AI, which is polarising for people who consider them as basic words in written and spoken language.
There is a chance this could become a big problem in the future, where some work will be turned down and people discriminated against for having certain verbal habits.