Artificial Intelligence | News | Insights | AiThority
[bsfp-cryptocurrency style=”widget-18″ align=”marquee” columns=”6″ coins=”selected” coins-count=”6″ coins-selected=”BTC,ETH,XRP,LTC,EOS,ADA,XLM,NEO,LTC,EOS,XEM,DASH,USDT,BNB,QTUM,XVG,ONT,ZEC,STEEM” currency=”USD” title=”Cryptocurrency Widget” show_title=”0″ icon=”” scheme=”light” bs-show-desktop=”1″ bs-show-tablet=”1″ bs-show-phone=”1″ custom-css-class=”” custom-id=”” css=”.vc_custom_1523079266073{margin-bottom: 0px !important;padding-top: 0px !important;padding-bottom: 0px !important;}”]

A Free, Independent ChatGPT Detector Reaches 90% Efficiency

Anti-plagiarism company Compilatio released an AI-generated text detector showcasing unprecedented reliability with both ChatGPT and YouChat.

French company Compilatio is addressing the growing concerns regarding the misuse of generative AI in education. Providing anti-plagiarism technologies to leading European universities, it just released a free, user-friendly AI detector. Preliminary tests show a 90% accuracy in attributing texts to either humans or artificial intelligence. Such performances rank Compilatio’s tool above its rivals. The detector works in several languages and reliably identifies content produced by ChatGPT as well as YouChat. Free for everyone to use and tinker with, it will serve as a testing platform for an upcoming commercial product.

Recommended AI News: LZG International, Inc. (FatBrain AI) Announces Plans to Uplist to OTCQB Venture Market

So far, Compilatio developers have tested their tool on a corpus of 125 texts authored by humans, ChatGPT, and YouChat. The detector could reliably label 100% of the content directly written in English by humans, as well as 100% of YouChat-generated works. It accurately attributed slightly more than 87% of ChatGPT texts.

“What sets our detector apart is how it accurately identifies not only AI-generated texts but also human-generated texts,” says Frédéric Agnès, CEO of Compilatio. “Most current systems are decent at identifying humans, but they also tend to misattribute AI-generated texts to humans, which somewhat defeats their purpose.”

Related Posts
1 of 35,313

Such test results are preliminary. By offering free access to its beta detector, Compilatio aims to conduct a larger-scale, more detailed assessment of its technology. Users can freely contribute by filling up a small survey every time they run an analysis.

Recommended AI News: Introducing WordPress VIP for Salesforce: a Powerful, Time-saving Integration for Content Marketers

“AIs are designed to produce contents that are as predictable as possible, following a principle akin to the word suggestions on your phone when you type an sms,” says Agnès. “These predictable answers result in a predictable vocabulary and syntax, which is precisely what our detector picks up.”

In the second half of 2023, Compilatio will release a commercial detector able to discriminate mixed AI and human contributions. It will highlight the suspicious passages and sentences and give an estimated probability for each of them. This add-on will integrate into the company’s anti-plagiarism software, providing a complete and updated solution for educational institutions. “Compilatio is now the only major player in the anti-plagiarism industry with an actual, efficient AI detector prototype. We are in a good position to lead this emerging market,” says Agnès.

Recommended AI News: Prisma Cloud Delivers Context-Aware Software Composition Analysis to Secure Deployment of Open Source Software

[To share your insights with us, please write to]

Comments are closed.