How to Grade Papers Written by #Artificial Intelligence  - 
The only answer to more tech is more human investment.  If we want our students to bring their human selves to the table, we have to create an educational environment that engenders human engagement.

https://rushkoff.medium.com/how-to-grade-papers-written-by-ai-4277a78a892
#For Teachers By Teachers #Edtech Innovation

How to Grade Papers Written by AI

Artificial Intelligence

For Teachers By Teachers

Edtech Innovation

How to Grade Papers Written by #Artificial Intelligence  -  The only answer to more tech is more human investment.  If we want our students to bring their human selves to the table, we have to create an educational environment that engenders human engagement.  https://rushkoff.medium.com/how-to-grade-papers-written-by-ai-4277a78a892 #For Teachers By Teachers #Edtech Innovation

Sam Mendoza

OpenAI has quietly discontinued its AI-writing detector, citing a "low rate of accuracy." The AI classifier, launched in January, was shut down on July 20. It was designed to help users differentiate between human-written and AI-generated text.
- OpenAI says the tool correctly labeled 26% of AI-generated text and mislabeled 9% of human-written text as AI-authored.
- The company plans to develop more effective methods for identifying AI-generated audio and visual content.
- AI-generated content detection has been a major concern in education, especially with tools like ChatGPT, which have raised worries about academic integrity.
- Researchers at the University of Kansas claim to have developed a method with over 99% accuracy to identify AI-generated scientific writing.
Meanwhile, a recent study showed that tools purporting to detect AI-generated text can be easily fooled.
- The study evaluated 14 detection tools, including Turnitin, GPT Zero, and Compilatio.
- While the tools accurately detected human-written text 96% of the time, the accuracy rate of ChatGPT text was only 74% and 42% for slightly tweaked ChatGPT content.

https://techcrunch.com/2023/07/25/openai-scuttles-ai-written-text-detector-over-low-rate-of-accuracy/

OpenAI has quietly discontinued its AI-writing detector, citing a "low rate of accuracy." The AI classifier, launched in January, was shut down on July 20. It was designed to help users differentiate between human-written and AI-generated text. - OpenAI says the tool correctly labeled 26% of AI-generated text and mislabeled 9% of human-written text as AI-authored. - The company plans to develop more effective methods for identifying AI-generated audio and visual content. - AI-generated content detection has been a major concern in education, especially with tools like ChatGPT, which have raised worries about academic integrity. - Researchers at the University of Kansas claim to have developed a method with over 99% accuracy to identify AI-generated scientific writing. Meanwhile, a recent study showed that tools purporting to detect AI-generated text can be easily fooled. - The study evaluated 14 detection tools, including Turnitin, GPT Zero, and Compilatio. - While the tools accurately detected human-written text 96% of the time, the accuracy rate of ChatGPT text was only 74% and 42% for slightly tweaked ChatGPT content.  https://techcrunch.com/2023/07/25/openai-scuttles-ai-written-text-detector-over-low-rate-of-accuracy/

Many perhaps unwisely relied on the tool to catch low-effort cheats, but OpenAI has retired it over its widely noted "low rate of accuracy."

OpenAI scuttles AI-written text detector over 'low rate of accuracy' | TechCrunch

A new study shows that tools purporting to detect AI-generated text can be easily fooled. The study evaluated 14 detection tools, including Turnitin, GPT Zero, and Compilatio, to identify words generated by ChatGPT. To bypass these detectors, students could simply tweak their essays and coursework, the results suggest.
- The researchers found that the tools, often marketed as anti-plagiarism platforms for teachers, had trouble detecting text that had been slightly rearranged by humans and altered using paraphrasing.
- While the tools accurately detected human-written text an average of 96% of the time, the detection accuracy rate of ChatGPT text was only 74% and noticeably lower — 42% — for slightly tweaked ChatGPT content.
- According to lead author Debora Weber-Wulff, the tools "don't work" and "don't do what they say they do."
- The study was published to the preprint server arXiv, so it has yet to be peer-reviewed.

https://www.technologyreview.com/2023/07/07/1075982/ai-text-detection-tools-are-really-easy-to-fool/

A new study shows that tools purporting to detect AI-generated text can be easily fooled. The study evaluated 14 detection tools, including Turnitin, GPT Zero, and Compilatio, to identify words generated by ChatGPT. To bypass these detectors, students could simply tweak their essays and coursework, the results suggest. - The researchers found that the tools, often marketed as anti-plagiarism platforms for teachers, had trouble detecting text that had been slightly rearranged by humans and altered using paraphrasing. - While the tools accurately detected human-written text an average of 96% of the time, the detection accuracy rate of ChatGPT text was only 74% and noticeably lower — 42% — for slightly tweaked ChatGPT content. - According to lead author Debora Weber-Wulff, the tools "don't work" and "don't do what they say they do." - The study was published to the preprint server arXiv, so it has yet to be peer-reviewed.  https://www.technologyreview.com/2023/07/07/1075982/ai-text-detection-tools-are-really-easy-to-fool/

A recent crop of AI systems claiming to detect AI-generated text perform poorly—and it doesn’t take much to get past them.

AI-text detection tools are really easy to fool

Finally, who understand what education is about.I have done my early studies until 18 in Europe and we had oral examinations to prove our knowledge and understanding of the topic. In the US, it was tests and essays. Sorry, but I learned nothing and passed with 3.9 GPA. It was way too easy and useless.

I never learned except during one-on-one discussions with professors about my essays. I never felt more lost than when my papers were read and graded by a teaching assistant that couldn't spare the time to give face-to-face feedback.

Finally, who understand what education is about.I have done my early studies until 18 in Europe and we had oral examinations to prove our knowledge and understanding of the topic. In the US, it was tests and essays. Sorry, but I learned nothing and passed with 3.9 GPA. It was way too easy and useless.  I never learned except during one-on-one discussions with professors about my essays. I never felt more lost than when my papers were read and graded by a teaching assistant that couldn't spare the time to give face-to-face feedback.

Britney

Cici Zhao

I agree, with one caveat. I’ve often had conversations with students who came across as intelligent, thoughtful and urbane whose essays were train wrecks. This bolsters your point in some ways, but it is disappointing to me that writing well is not equal to “being a good thinker.” Also, I suspect my good talkers have a high EQ and this method of evaluation would simply penalize students with a low EQ, some of whom might be good writers.

Sam Mendoza

A group of likeminded people in For Teachers By Teachers are talking about this.