Dear Commons Community,

Leon Furze has an article, entitled “Ditch the Detectors: Six Ways to Rethink Assessment for Generative Artificial Intelligence” that scales assessment from active use of AI to doing all assessment in-person. Here is a summary of his recommendations.

• Use Level 5 assessments – At this level, teachers actively encourage students to experiment with generative artificial intelligence (GenAI) tools like ChatGPT. “When students leave the educational bubble,” says Furze, “they’ll be free to use whatever tools are available to them.” This gives them practice, with monitoring and accountability, using any applications suitable to the task – text, image, audio, video, 3D, and generating code. Drawbacks include ethical concerns (copyright and intellectual property) and equity of access to GenAI apps.
• Expecting AI use and teaching the skills – It’s realistic to assume that most students will be using AI tools, so it’s smart to teach them how to use them for ideation, editing, and appropriate portions of class assignments. Since everything is in the open, this eliminates the need to use detection tools, and teachers can address students’ concerns and knowledge gaps. Drawbacks include the time, resources, and cost required for educators to get up to speed on the technology, as well as the need to update and reframe current assessment tasks.
• Ungrading – “If we shift the focus of education away from the final assessment and towards what is being taught (and why),” says Furze, “then the imperative for academic misconduct may be lessened” and students may focus more on deeper understanding and genuine learning. Grades keep students focused on GPAs and transcripts versus growth and improvement, says Emily Pitts Donahoe. De-emphasizing grades reduces stress and pressure and allows for imaginative use of GenAI without worrying about the impact on final grades. Drawbacks include going against the grain of many schools’ assessment practices and the notion that ungrading won’t work in “real world” subjects.
• Knowing students’ style and voice – Whether through high-tech tools (“stylometry”) or just plain “knowing your students,” this is what some teachers are doing to guard against inappropriate use of GenAI tools. The advantages include building relationships with students and understanding and appreciating their perspectives and ways of expressing themselves – and strong relationships can help prevent academic misconduct. Drawbacks include whether this can be scaled beyond small classes and whether it really stops the most sophisticated forms of cheating – there are tools that can emulate a student’s style.
• Redefining cheating– This strategy, says Furze, raises the fundamental question: how do we assure that learning has happened? It potentially “allows us to approach academic integrity from a more-constructive standpoint, emphasizing the importance of genuine learning over the moralistic labelling of certain behaviors. By moving away from punitive measures and instead designing assessments that truly demonstrate learning, we can create a system that encourages students to engage with their education meaningfully, rather than seeing it as a series of hoops to jump through.” Redefining cheating also “demonstrates to students that we value trust and transparency and places the expectation on them to do the right thing.” And it has the additional advantage of reducing educators’ workload.
• In-person, in-time, in-place assessments– “There are plenty of methods that predate GenAI by a few centuries,” says Furze, “and still work.” This doesn’t mean examination-style assessments; it includes group work, orals, seminars, simulations, brainstorming with sticky notes, debates, marker pens on butcher paper, and more. Advantages include that these assessments are easy to monitor and secure, since students don’t have access to devices and can be relevant, engaging, and authentic. Drawbacks are that this kind of assessment is difficult to scale for large classes, and there’s no online option.

Good practical suggestions.  I started incorporating Level 5 assessments using generative AI last year.

Tony

This entry is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.