Penn State begins ‘cheat-a-thon’ to test limits of generative AI
The Center for Socially Responsible Artificial Intelligence at the The Pennsylvania State University will host a “cheat-a-thon” competition starting Monday that will explore the benefits and risks of the current wave of artificial intelligence technologies.
The online event, which runs from Monday through April 6, encourages all students with .edu email addresses from around the country to participate in a competition designed to stretch the capabilities of generative AI models like those developed by OpenAI or Anthropic.
The competition has two phases. Phase one includes collecting questions that university faculty believe will be difficult for generative AI tools to answer. The second phase involves students attempting to answer questions only using generative AI and without the aid of “online libraries, search engines or Wikipedia,” according to the Penn State website.
The university plans to award cash prizes to students who score the highest grades based on the answers they submit to the questions.
“Generative AI tools are making tasks faster and easier to complete, but they can also amplify inequalities, spread misinformation and present serious risks,” the competition announcement reads. “Identifying these strengths and dangers can help university faculty and students use generative AI in safe and ethical ways without compromising academic integrity.”
