Just before Christmas, the internet was alight with various pieces
on ChatGPT, an artificial intelligence (AI) chatbot which is able to generate
written pieces/articles, generate code and compose songs and poems in various styles. Users input a small
list of items, and ChatGPT produces the goods. The AI learns continually through Reinforcement Learning with Human Feedback (ELHF).
The platform is free (for the moment) with over a million users
signed up within hours of launch, leading to it being often unavailable due to
the sheer number of people using it. A larger number of information summaries,
blogs, opinion pieces (some written by ChatGPT) have flooded the internet. These users, help the AI learn through its interactions, helping to increase the authenticity and realism of the outputs.
Relevant articles include ones related to how AI disrupts industries, its effect on education / classrooms, teaching practice and learning, how it could transform learning, the higher education response, including a 'if your can't beat them, join them idea '.
As usual, there is no available response contextualised to vocational education (sigh).The types of jobs threatened by AI bots include copy/content writing, customer support, and software developers /programmers, although the general consensus is that these jobs are safe for the moment, given the germinal nature of ChatGPT. Given the AI tends to provide written responses, similar challenges to those brought up in the articles above within school and higher education contexts arise when ChatGPT is used to develop written material for vocational education based assessments. For example, project proposals and the ensuing report, can be generated when an outline is presented to ChatGPT. Thankfully, practice-based assessments are the mainstay of vocational education and the learner still has to perform the required skills to the expected standards.
How can vocational learners be prepared for the use of increasingly powerful AI tools that generate images and text when provided with the right cues? To begin, it is important for learners to understand the strengths and weaknesses of these platforms. Learners also need to provide the chatbot with the right information and the right occupational jargon, to ensure the output generated is realistic. This requires learners to have a good grasp of the content/topic, before they feed in the cues that are used to generate the output. The mechanics of writing (grammar, spelling, sentence structure etc etc) may be taken care of but it is still up to the user, to check, evaluate and finalise the resulting piece of writing.
As someone who spent many years honing the craft of becoming an academic writer, I can see the usefulness of something like ChatGPT to generate starter paragraphs. These emergent pieces may be used as precursory summaries, to help overcome 'writers' block' or to provide some of the scaffolds for the article's argument. Having tried out the platform with several 'article skeletons' I have found the generated text to be repetitive and sometimes long-winded. However, this could be due to the cues I have provided! It also works from what the user provides, so if the cues are not factually correct, the platform will be none the wiser.
ChatGPT represents a challenge to how young people are taught how to write when written texts can be generated. Schools could ban the platform as it reduces the writing of school reports and essays, to plugging in some items around a topic, generating the text and copy and pasting. The long journey of learning how to write through practice, trial and error, and continued improvement, becomes lost when the act of actually consolidating a string of items into a cohesive whole is lost :( Therefore assessments focused on essays will need rethinking. It is now more important than ever, to think of assessments as FOR learning rather than OF learning. ePortfolios are a means to record the learning / writing journey and assessments based on evidence of 'reflective learning' rather than the end result (essay, report etc.) are one way to work around the challenges of AI generated work.
We also need to think through the implications, ethics and philosophical underpinnings, When AI is able to respond in ways so similar to humans, it becomes difficult to tell the difference. As humans, we might lose some of our humanity, when a machine 'takes over' thinking and creative writing from us. With it, goes the self-actualisation associated with positive feedback from putting effort into difficult tasks. Evaluating the work of others, is not a replacement from putting in the hard work required to bring a creative piece of work into being. It will be difficult to push against the tide of AI generated work. We need to think through the wide implications, one of which may be an increased valuing of artisanal work, the one off bespoke products. AI may design these objects, 3D printers may be able to produce by the millions, but 'one of a kind' human designed and crafted products may be one way to ameliorate the rise of 'non-human' designed and manufactured goods. Vocational education, especially in the trades, will need to therefore emphasis quality of skills, to meet the market for bespoke items.