What can GenAI tools do well?
This activity explores some capabilities and limitations of generative AI in the context of higher education teaching and learning.
It contributes to your achievement of the course learning outcome: describe the key capabilities, limitations, challenges and opportunities of generative AI in higher education teaching, learning, feedback and assessment. More specifically, by the end of this activity, you will be able to:
- Evaluate key capabilities and limitations of generative AI tools.
- Describe some ways to get the best outputs from generative AI.
- List ways students are already using generative AI tools.
In this step, you will hear from Martin about some of the capabilities of AI tools and, in the text below, you can read about some very specific ways he uses these tools to support his work.
I use Bing Chat, Google Bard and ChatGPT routinely in my academic work after working through several initial frustrations. Likewise, after much time fiddling, I have found ways to exploit the potential of AI image generators such as Dall-e, stable diffusion and Midjourney. My top five uses all relate to work that is essential but can be time consuming:
- Taking automated video transcripts, which is in itself an AI technology (eg from YouTube), and then using a text generative AI tool to punctuate and correct the transcript. I can then use the transcript to write a video summary.
- Illustrating resources and presentations with images created using generative AI tools.
- Turning my notes and rough jottings into coherent and shareable notes that are accurately spelt and logically organised.
- Creating scenarios and vignettes for teaching. I offer a structured framework and then ask the tool to create short outputs for in-class discussion.
- Creating lists of terminology with definitions, which I can then use in other tools to create self-access flashcards or as the basis for multiple choice question authoring for in-class use.
The challenges presented and the worries about any tool’s apparent ability to pass any exam should not be ignored, of course. Lecturers and teachers would be wise to see how readily available AI tools perform when confronted with questions and tasks that are similar to those they set for their students. Then, educators can consider how assessments may be modified. Students need to ensure they are alert to local regulation and institutional policies as a first step, and that they understand that it is never a good idea to depend entirely on automated output, as impressive as it may be.
Critical engagement with credible sources cannot be replaced by use of generative AI tools but these tools can help us access, process and work with those credible sources. The most important consideration when it comes to capabilities is to focus on the ‘assistance’ aspect – what is it that you need to do that might be improved and/or sped up with the assistance of a generative AI tool? If you attempt to offload cognitive processes you may find the outputs are disappointing, not to mention the impact on learning!
When you have completed this step, you will have seen that to employ AI tools successfully you need to critically engage with whatever subject matter you are working with and with the tools you employ. GenAI is a powerful assistant but it cannot and should not be used in an attempt to replace cognitive engagement. In the next step, you will learn how to write effective prompts to generate useful content from GenAI tools.
Additional resources
If you would like to explore this topic further, here are some additional resources.
The videos in this playlist, Generative AI Practicals, have all been through the process described in point 1 above. The time it saved me is considerable on every occasion.
An example of point 2 above: this post, The Ideal Learning Environment, exploited multiple forms of AI in its creation (explained in the text) and includes two AI-generated images.
This short post and video, Generative AI practicals: making sense of lecture notes (with ChatGPT), shows a worked example of point 3 above.
This post about generating micro case studies illustrates the approach from point 4 above.
In terms of point 5, I created this set of flashcards using ChatGPT, which came up with the initial list and definitions separated by a comma, and I then transferred these into ‘Quizlet’ to create a study set. Quizlet uses built-in adaptive learning AI too.
News article on ChatGPT’s ability to ‘pass’ exams.
Join the conversation
If you are a user of generative AI already, what are your top uses? If not, what might you use these tools for?
Reach your personal and professional goals
Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.
Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.
