OpenAI can't tell if something was written by AI after all

OpenAI shut down a tool that used AI to tell human writing due to low accuracy rates. In an (updated) blog, OpenAI said it has decided to phase out its AI classifiers by July 20. "We are working to incorporate the feedback and are currently researching more efficient generation techniques for text," the company said.

As it launches the tool for capturing AI-generated writing, OpenAI said it plans to "develop and deploy mechanisms that enable users to understand whether audio or visual content is AI-generated." generated or not. generated or not. However, there's no word yet on what those mechanisms might be.

OpenAI fully acknowledged that the classifier was never very good at catching AI-generated text and warned that it could lead to false positives, i.e. human-written text being tagged as AI-generated. Could , OpenAI shut down the tool before adding the update, saying that the classifier could improve with more data.

After OpenAI's ChatGPT entered the landscape and became one of the fastest growing apps ever, there was a race to understand the technology. Several sectors raised concerns about AI-generated text and art, particularly teachers who worried that students would no longer study and let ChatGPT write their homework. New York schools banned access to ChatGPT on school grounds amid concerns about accuracy, security and cheating.

Misinformation via AI has also been a concern, with studies showing that AI-generated text, such as tweets, can be more reliable than text written by humans. Governments have yet to figure out how to rein in AI and so far, they are leaving it up to individual groups and organizations to set their own rules and regulations to deal with the onslaught of computer-generated text . Develop your own protective measures. And it seems that for the time being, no one, not even the company that helped start the generic AI craze in the first place, has any answers about how to deal with it all. , Although some have caught on, it is becoming increasingly difficult to easily distinguish between AI and human action.

OpenAI has also recently lost its trust and security leader, at a time when the Federal Trade Commission is investigating OpenAI over how it handles information and data. OpenAI declined to comment beyond its blog post.