Generative artificial intelligence technologies (e.g., Dall-E, ChatGPT) use machine learning algorithms to collect data, monitor actions and interactions and respond to further inputs and feedback in order to fine tune its engines. Many of us have been using AI-based tools, such as Google Maps, predictive text or Grammarly, in everyday life for years.

The quality and validity of outputs from AI tools such as ChatGPT can be highly variable, depending on the prompt it has been given and the way the algorithms have been trained. Academic users have reported examples of these tools falsifying references and making up quotesLink opens in a new window which aren’t always easily detectable and clearly compromise any work using this content. AI tools have also been criticised for creating overly generalised responses.

Used thoughtfully, AI tools can be used to support academic practices:

• They can be used for academic writing in order to allow users to support the flow and structure of writing, alongside more familiar tools like spelling and grammar checkers.
• A well-crafted prompt can generate basic content that can be refined and corrected (where necessary), allowing learners to focus on higher-order aspects of learning and exploring concepts.
• They can support student learning, for example, they can explain computer code in plain EnglishLink opens in a new window, support language conversation practiceLink opens in a new window and generate images in response to text prompts.

Generative AI tools will continue to impact and influence the way that we engage in learning and teaching in higher education, in similar ways to previous technologies such as spell-checkers, online libraries and calculators, as well as in new and unknown ways.

It is not practical or beneficial to try to ignore, ban or eradicate AI. At Warwick, the preferred path is to work on developing assessment, teaching and learning in ways that promote responsible engagement with AI. As the dust settles and staff move beyond early responses to the rapid improvement of AI technologies, it will be important to carefully consider the purposes of assessments and the forms of evidence of learning that are valued in relation to modules, units and students.

The emergence of generative AI technologies like ChatGPT highlights already existing questions about how we can refine assessment in relation to what matters to staff, students, and society more broadly. It prompts us to consider alternative assessment options, including those that most closely align with educational principles promoted at Warwick. Pedagogical approaches that value assessment that promote students' understanding, and that is meaningful and valuable in its own right (see our guidance Academic Integrity) continue to be relevant in this changing context, and may help staff and students to navigate some of the challenges that are emerging.

By proceeding knowledgeably, thoughtfully and ethically in all aspects of learning, teaching and assessment, we will be well placed to navigate the challenges and opportunities of a dynamic technological landscape.

There are also some pressing concerns and practical challenges that apply to existing approaches. For example:

• What immediate changes are needed to the way assessment is designed and conducted?
• How do the rules and principles of academic integrity apply in relation to ChatGPT or other technologies?

Some important actions to mitigate the immediate risks posed by generative artificial intelligence include:

1. Learn about how the technologies work, as well as their benefits and limitations.
2. Engage in meaningful conversations with students about the ethical considerations of emerging technologies.
3. Explore where and how academic principles, concepts and rules might need to be adapted to incorporate generative artificial intelligence.
4. Review assessment designs, policy, guidance and communication around acceptable and responsible use of AI technologies in units and courses.