GPTZero

The second-largest teacher's union in the U.S. has partnered with a company that can detect when students use artificial intelligence to do their homework.The American Federation of Teachers (AFT) recently signed a deal with GPTZero, an AI identification platform that makes tools that can identify ChatGPT and other AI-generated content, to help educators rein in, or at least keep tabs on students' reliance on the new tech."ChatGPT can be a really important supplement and complement to educators if the guardrails are in place," AFT president Randi Weingarten told CBS MoneyWatch. "And the guardrails have to be about privacy and security and things like that."Working with AI, not against itThere is, without a doubt, a place for AI in the classroom, according to Weingarten."We believe in its potential and we know if we don't guard against its perils upfront, we're going to repeat the terrible transitions that happened with the industrial revolution," she said. Products like those provided by GPTZero will help educators work with and not against generative AI, to the benefit of both students and teachers, in Weingarten's view. "You can't stop technology and innovation. You need to ride it and harness it and that's what we are talking to our members about," she said. GPTZero, a 15-person company co-founded by recent Princeton graduate Edward Tian, has developed tools for people in the front and back of classrooms. "We're committed together to figuring out the applications of AI in classrooms, and building GPTZero to be the best pedagogical solution for teachers and students to collaborate together in adopting AI," Tian told CBS MoneyWatch.Free versions of GPTZero products are available. The teacher's union is paying for access to more tailored AI detection and certification tools and assistance. Using AI responsiblyDeveloped in January to scan text for AI input, GPTZero has since launched new tools, including one that allows students to certify their content as human, and to openly disclose when they use AI. "A big goal of ours is to demonstrate that the use of AI in education does not have to be adversarial," Tian said. "In January when everything was starting, there was the mentality that it was taking the plagiarism model of copying and pasting content, which is not the right framework here."Ultimately, said Tian, he wants to help teachers and students work together to make the most of cutting-edge AI technologies while mitigating their potential to do harm. "We are working with teachers to figure out where AI fits into education. We want to empower students to use AI responsibly," Tian said.  Weingarten sees upsides to AI for teachers, too. For one, she said educators aren't Luddites and are already adept at using tech tools in classrooms. "It can hugely reduce paperwork burdens, bureaucratic burdens, and it can help with the writing of lesson plans," she said of AI technology. "I think there is huge potential here, but we have to be sober about it. We cannot pretend that it is a panacea, but have to hope and push for the kind of ethical regulations that are necessary so that it doesn't destroy."

The findings reveal a clear bias in GPTZero, which tends to assign lower AI-generated probabilities to AI-assisted abstracts written by native authors (median = 9.50%, mean = 30.68%, SD = 35.29%) and higher probabilities to those written by non-native authors (median = 22.50%, mean = 44.61%, SD = 43.33%). Welch’s t-test confirms this disparity, showing a statistically significant difference (t = −2.115, p = 0.036). This pattern, visualized in the density plot in Fig. 4, highlights a concerning imbalance that potentially places non-native authors at a serious disadvantage, amplifying existing inequities in scholarly publishing. In contrast, the other two tools, ZeroGPT and DetectGPT, exhibit more balanced performance, as evidenced by their Welch’s t-test results, which show no statistically significant differences (p = 0.687 and p = 0.607, respectively). However, given GPTZero’s high accuracy in the previous scenario, these findings raise significant concerns about the unintended consequences of its use.

In response to these concerns, Tian emphasizes that GPTZero is moving away from a “plagiarism mentality” toward promoting mindful A.I. usage.

On January 2nd, Tian released his app. He named it GPTZero. It basically uses ChatGPT against itself, checking whether "there's zero involvement or a lot of involvement" of the AI system in creating a given text.

After ChatGPT came out, I was inspired to investigate where and when AI was being used online. It took about a week over winter break to build the first version of our AI detection tool, and it was released in January 2023.

Edward Tian, who wrote the app that sniffs out ChatGPT's bot-written text, said he will keep a version of ZeroGPT free, even as he is being approached to work with the plagiarism-detection service Turnitin and other companies.

When we first started GPTZero, I think I was actually more pro-AI than I am now. I believe people should know when something is AI-generated, but I wasn't saying not to outsource anything at all.

Three years later, we sold our company, GPTZero, to Superhuman, and have stayed on board to work with the team. Superhuman is a very pro-AI company, and I think it was an unexpected move for us as an AI detection company to join them.

I built an AI detection tool in college. It exploded in popularity and we sold it — now I feel sad for the internet at large.

I've started to see a friction point at work when people can't write or edit, and everything they send reads like AI. Part of what I'm doing now at Superhuman is working on responsible AI usage policies and guidance.

In kicking off this initiative, GPTZero partnered with Greg Altman, an Emmy-nominated TV producer; Bill Retherford, a journalist and 4-time Emmy winning documentarian; and Wendy Snyder, a college admission officer at Northeastern University. All three created “digital doubles” that they helped prompt and calibrate and can provide the kind of feedback they would’ve offered had they actually read your work.

In the past year, though, I've started to see an over-reliance on AI among people who finished university after ChatGPT was launched, especially with writing. If I went back to college right now, I wouldn't major in computer science. I'd probably major in journalism and get a minor in computer science. I'd get a humanities degree because those are the important skills now.

We created artificial scarcity to help grow the business

The bigger mission we have is to take steps to ensure people are being respectful of the person who's on the receiving end of AI generations and productivity now. We want it to be that we're reaping the benefits of AI without making other people deal with more work.