While some teachers love the ease with which they can generate a lesson plan or create a set of questions based on a text, others are finding the time they need to spend adapting and ‘tweeking’ the end result leads to a dubious net benefit in terms of time saved.
Authors and content creators too are assessing how AI can fit into their broader work plans, amid concerns that some generative AI models are developed off the back of pirated work stored in massive online databases such as LibGen.
Some content creators are now calling for payment systems to be developed which recognise the importance of paying the original creator of a work for their content.
They argue that as big tech companies develop mass models of content, they should be paying for the content they use.
In contrast, the counter argument runs that using content without permission is within the bounds of ‘fair use’ which allows limited use of copyrighted material in some circumstances.
Other creators are pushing back against the concept that AI tools can be trained using their content, particularly if it has been accessed through pirate databases.
Essentially it has become a battle between protection of creative works and the people who created them and the ability of companies to utilise content they acquire for their own benefit.
So how does this apply to the world of teaching and learning? Some teachers revel in their ability to rapidly generate content, check information, find websites and make visual images.
They love being able to use a tool such as Magic AI A recent study has found that, far from being a time saver, AI can in fact create work for teachers.
The process of developing strong, effective prompts and questions takes time. So too does adapting the end result to cater for local conditions and curriculum requirements.
Many teachers have discovered that, although AI tools can be effective in generating a set of questions for example, they struggle to design questions which are varied and engaging.
An initial prompt centred around a key topic or chapter from a text can result in a question set which aligns precisely with the requirements – sometimes a little too precisely.
As a result, students can get bored and frustrated with the repetition, or simply turn to AI tools themselves to respond.
Another challenge created by AI tools for content generation is the introduction of bias. Some teachers and researchers have noted the tendency to reinforce stereotypes when creating scenarios or case studies, such as characters who excel at maths and science being more likely to have male names, while characters who display compassion and care are more likely to be female.
Cultural bias has also been reported, with some researchers noting that AI can tend to reinforce stereotypes or simply avoid cultural references unless they are clearly requested in the prompt.
One of the skills an experienced teacher brings to resource and lesson plan development is the ability to pay attention to the changing needs, interests and focus of their students.
They are able to differentiate to match content to the point of challenge, and identify resources which are locally relevant and engaging.
They can also make judgements about the relative merits of authors and publishers, relying more heavily on evidence based, reputable sources and less on blog posts, forums, discussion pages and online reviews.
AI draws equally on sources of data it finds online unless given specific prompts to use particular evidence sources, relying mostly on finding the closest match between the text of a question and the corresponding answer it generates.
For many teachers, AI ultimately represents a potential time saving, vital in the context of long work hours and excessive administrative requirements that eat into lesson planning time.
It does come with its caveats however, and only time will tell if the benefits of using a system developed with dubious origins will morph into a reliable, reputable tool that meets the needs of teachers and their students in Australian schools.
