|
AI is already widely used across industries, including healthcare, finance, IT, entertainment and education. |
| Your time at the University of Exeter will enable you to understand the limitations and risks of AI technologies, evaluate AI-generated outputs, identify biases, and understand the unique value you bring to the table. |
GenAI tools can be useful for many things, including:
But to make the most of these benefits of AI, we need to have an awareness of its limitations and how to use it responsibly. Let's quickly review how AI works to help us understand the limitations we're going to look at.
As we can see, GenAI works by making predictions based on data it is given; it doesn't 'know' anything. This means it has various limitations. Let's explore some of these!
Before relying on AI-generated content, ask yourself:
Is the information accurate and reliable?
Does it align with trusted academic sources?
Have I critically analysed the output, or am I relying too heavily on AI?
If you answer 'no' to any of these questions, you need to re-think whether it's appropriate to use AI for that particular task.
If you do decide to generate AI outputs, you can thinking critically about them by applying the CRAAP checklist:
|
Currency |
Is the information up to date? |
|
AI Context: GenAI tools may not have access to real-time data or current events unless connected to the web. Check the date of the information and verify with current sources. |
|
|
Relevance |
Does the information meet your needs? |
|
AI Context: AI may generate generic or off-topic responses. Ensure the output is tailored to your question, discipline, and academic level. |
|
|
Authority |
Who is the source of the information? |
|
AI Context: AI often doesn’t cite sources unless prompted. Check whether the response references credible authors, institutions, or publications. If not, verify externally. |
|
|
Accuracy |
Is the information correct and supported by evidence? |
|
AI Context: AI can make up facts. Cross-check claims, statistics, and quotes with trusted sources. Be cautious of confident-sounding but incorrect answers. |
|
|
Purpose |
Why was this information created? |
|
AI Context: AI outputs are generated based on patterns, not knowledge. Consider whether the response is informative, persuasive, biased, or misleading. Be aware of potential bias in training data. |
Optional activity:
1. The below images were created using GenAI tools and the prompt 'Realistic image of student life at the University of Exeter, on an average day, in a photographic style’. Answer the following questions:
How accurate are these outputs? Can you spot any errors or biases?
Why might these inaccuracies and biases be present in these outputs?
How would you improve the outputs based on your knowledge and experience?
2. Use a GenAI tool to generate a text summary of student life at the University of Exeter, and consider the same 3 questions above.
3. Reflect on this activity by answering the following questions:
What assumptions do you make/have you made about GenAI tool outputs?
How will this exercise influence how you use GenAI tools in your future education? In your future career?
Image generated using DALL-E in ChatGPT (Premium) from the prompt ’realistic image of student life at the University of Exeter, on an average day, in a photographic style‘.
Image generated using Imagen 3 Google Gemini (Advanced) from the prompt ‘realistic image of student life at the University of Exeter, on an average day, in a photographic style’.
Want to learn more? Register for the 'Gen AI Under the Microscope: Thinking critically about it' workshop.
University of Exeter LibGuide is licensed under CC BY 4.0
