OpenAI Shuts Down AI Text Detector Due to Low Rate of Accuracy

Less than six months after OpenAI launched its AI text detection tool, the company has decided to quietly pull it back due to inaccurate results.

OpenAI edited its January announcement of the AI classifier tool, adding a statement that - As of July 20, 2023, the AI classifier is no longer available due to its low rate of accuracy".

The company went on to add that it's working on incorporating feedback and researching for more effective provenance techniques for the text". The statement also said that OpenAI is committed to developing and deploying mechanisms that allow users to understand if any visual or audio content is AI-generated.

OpenAI's AI Text Detector Was Imperfect, the Company Warned All Along

OpenAI rolled out the AI text detector following rising concerns over the misuse of ChatGPT and similar generative AI services by students to write essays and complete homework. The free-to-use tool was quite simple and required users to copy and paste the necessary text into it.

The tool then detected whether the piece of text, be it a blog, an email, or an essay, was AI-generated or written by a human. The AI text classifier was powered by a large language model that ranked how much of the text was potentially AI-generated, from very likely" to unclear" to likely".

However, even on the day the tool was launched, OpenAI announced that it wasn't entirely perfect and the results should be taken with a grain of salt.

We really don't recommend taking this tool in isolation because we know that it can be wrong and will be wrong at times - much like using AI for any kind of assessment purposes.Lama Ahmad, OpenAI's policy research directorr

He added that educators must be very careful about how they include the tool in making decisions about academic dishonesty.

In other words, the tool was only supposed to be another reference point, such as comparing a student's work with their past assignments and usual writing style..

Back in May, an instructor at the University Of Texas A&M-Commerce came under fire for withholding some students' grades after their text was predicted to be AI-generated by ChatGPT. Following the backlash, the university decided to investigate the issue and reinstated the students' grades.

What Are the Implications?

While this might be the first major AI text detection tool to be shut down due to a lack of accuracy, OpenAI's AI detector is far from being the only AI classifier to suffer from this issue.

Detecting AI-generated content is incredibly challenging, and similar tools developed by other companies have often been known to be unreliable.

While the inaccurate detection of human-written content as AI-generated text can potentially have severe implications for students, the lack of a reliable detection tool also enables cheating.

The growing popularity of ChatGPT raised the alarm among educators, who are worried that it would be even easier for students to cheat by getting AI to do all the work.

However, some of the alternative AI classifier tools in the market are known to be more accurate. The AI software rolled out by Turnitin claims to tackle plagiarism with 98 percent confidence" and has been adopted at various schools and universities. Other competitors like ZeroGPT have been gaining popularity too.

The post OpenAI Shuts Down AI Text Detector Due to Low Rate of Accuracy appeared first on The Tech Report.