A recent study by researchers from the University of Stanford (California, USA) has demonstrated an amazing phenomenon concerning the use of LLM (Large Language Model – Large model of language) like that of Chatgpt. Indeed, if one might think that the adoption of such artificial intelligence technology could be done very quickly in areas particularly equipped and educated in the use of this technology, it would seem that this is not the case with regard to the figures.
A fact that materializes even more by the words of researchers from the University of Stanford. “” “Our study shows the emergence of a new reality in which firms, consumers and even international organizations are based on generative AI for their communications“They said in comments reported by Ars Technica.
Indeed, to come to this conclusion, an in -depth analysis on the adoption of LLM in different sectors, between January 2022 and September 2024, was carried out.
And the figures give the dizzy a little. In total, 687,241 consumer complaints, 537,413 corporate press releases, 304.3 million job offers and 15,919 United Nations press releases were screened.
Using a detection system highlighting artificial intelligence assistance in the production of a document Researchers have shown that:
- 18% of consumer complaints,
- 24% of corporate press releases
- 15% of job offers
- And 14% of the United Nations press releases had received assistance from artificial intelligence in their writing.
But it was not these figures that have most surprised the researchers, even if they are impressive. Indeed, scientists were surprised to note that if the adoption of this technology in the United States was, overall, higher in so-called “educated” areas, the frequency of use of these new tools, like Chatgpt, was much higher in so-called “less educated” areas.
An “equalizing tool” that challenges prejudices?
Even if it may seem surprising that the United Nations uses artificial intelligence to write their press releases, this is not what interlocked researchers.
Indeed, by putting aside press releases and job offers, special attention was paid to consumer complaints.
Allowing to have visibility on the area from which a document is issued to a particular service, the researchers were therefore able to highlight that proportionally, certain states where the population is considered “less educated” had adopted artificial intelligence at a higher rate than in other states, more populated, but also considered to be “more educated”.
Thus, in Arkansas, out of 7,376 consumer complaints, 29.2% had been written with the help of the AI. In the Missouri, out of 16,807 complaints, 26.9% had been written using artificial intelligence. Finally, the North Dakota completes the podium where on a little more than 1000 complaints, 24.6% had been produced, entirely or in part, using an LLM.
In contrastStates concentrating a large population and access to technology much easier and democratized such as California (17.4% of 157,056 complaints) or New York State (16.6% of 104,862 complaints) showed much lower AI adoption rates in the production of these documents.
An important observation which allowed researchers to express the theory of artificial intelligence as an equalizing tool between rural and urban areas.
“” “While the digital divide between cities and the countryside seems to persist, our observation that low -level regions have slightly higher LLM adoption rates in consumer complaints suggests that these tools can serve as equalization tools in consumer defense.”They explained in comments reported by Ars Technica.
However, even if the figures seem to speak of themselves, the study is limited only to content (complaints, press releases, job offers) written in English and disseminated online.
But more importantly, although the researchers were able to detect AI assistance in the production of documents, they explained that they were not in the ability to detect the possible modifications made by a human being on a text generated by an AI, but also to detect the texts that have been specially written by coded AI to perfectly imitate a human writing style.
In fact, future studies will have to be carried out to find out whether or not AI is really an equalizing tool in society.
Source : Ars Technica
With an unwavering passion for local news, Christopher leads our editorial team with integrity and dedication. With over 20 years’ experience, he is the backbone of Wouldsayso, ensuring that we stay true to our mission to inform.
