Is GenAI an Existential Threat?

The initial response from the higher education sector was driven by anxiety, with concerns about cheating and threats to academic integrity. As the debate has evolved, there has been greater consideration of the opportunities GenAI presents and a recognition that the effective use of GenAI will be a skillset all students will require in the future. 

Nevertheless, important questions remain for educators, especially regarding assessment design where approaches that seek to test content knowledge and not application are prone to the risk of GenAI. As a result, there has been renewed interest in alternative approaches such as authentic assessments and ungrading. 

How Can GenAI Enhance Student Learning Experiences?

• Creating personalised learning materials and pathways that tailor content to meet the diverse needs and learning preferences of students. 
• Automating administrative tasks such as lesson planning and grading, leaving more time for educators to provide personalised and timely support and feedback. 
• Analysing student data to identify patterns and trends to inform future teaching strategies. 
• Creating flipped learning experiences: by asking students to analyse AI-generated resources for accuracy, relevance and effectiveness, thereby increasing appreciation of AI and its limitations.

Limitations of GenAI Tools

While GenAI has immense potential, it’s also a work in progress and there are many issues that, as educators, we need to mitigate against.

• Hallucinations are where GenAI platforms write statements that sound confident but are not necessarily backed up by reliable data.
• Bias is a known challenge, since the data on which many GenAI platforms have been trained is predominantly white, Western, and English. As a result, some GenAI outputs, such as images of people in particular professions, reinforce common stereotypes.
• Currency is an issue as GenAI is not connected to live sources of information like the internet but is based on data supplied at a particular point in time. For example, many GenAI platforms struggle to take account of the impact of Covid-19 as the training data predates the pandemic.
• GenAI is often unable to cite sources, where these sources sit behind paywalls, or training data is incomplete. This can also result in misattribution of content.
• Personal data used with GenAI can be a breach of privacy leading to individual and institutional liability.

Student Assessments and GenAI use

Assessment has been at the frontline of the battle with GenAI because of the technology’s capacity to enable cheating. But we must remember that the majority of students don’t want to cheat, but that external pressures may lead them to make poor decisions. There are things we can do to avoid this:

• State what your AI policy is, ideally by working with students to develop a shared position.
• Invite students to be transparent about how they have used AI and to submit a reflection as a means of getting students to engage with ethical considerations. Consider involving students in the assessment process.
• Consider Ungrading/Grading for Growth as an alternative that reduces the pressure to cheat.
• Change your assessment format to be more formative and authentic or to incorporate more personal reflections.
• Use project-based assessments or term-long community action projects and grade for quality of contribution as well as quality of output.