AI exploration
When looking at new tools, it can often be useful to experiment with them to find the approaches that work for your requirements.
Testing a variety of tools can be useful as each AI tool works differently and usually has a particular type of output that it can create (text, image, media etc.). This will enable you to find the right tools that work for you.
It is essential to think about the data you are inputting, even when conducting informal tests. If you are planning to explore the use of AI tools and how it will work in conjunction with your university studies, we would recommend thinking about the concepts below.
Data
Think about the data or prompts you are using with AI systems as this could potentially become data that is shared with wider groups. The more suitable prompts will be based around exploring ideas, asking questions, and generating new content. However, it is not recommended that you:
- Put any form of personal data (your own, or the data of others) into AI systems as this would be considered a GDPR breach.
- Copy and paste work into the tool to check/alter/improve it as this will be considered an academic offence and could be sharing your personal data.
Analyse elements with a critical eye
One of the limitations of AI (web) is that it can produce false information (hallucinations). When experimenting, especially if asking it to collate research or answer a question, it should be examined closely before use to ensure the data is accurate.
Identifying the source of the data it has used can be a useful way of determining if the information is from a trusted resource. AI sometimes produces false references or sources that appear correct, so be sure to verify each one. For more help with this see the AI Study Skills Guide.
Identify and experiment with prompts (tone, detail, approach)
As mentioned previously, different AI tools respond differently to prompts; it can be useful when experimenting to think carefully about the prompts given to the system. Simple variations in length, tone, and approach can have differing outputs. Sometimes it may take several attempts at reworking prompts before you receive an output that is appropriate.
Define an output.
When designing a prompt, be very clear with the output that you are looking for, especially if you would like it in a certain format or type. This clarity will help the AI to support you when it’s gathering information.