Since the launch of Open AI’s ChatGPT preview at the end of last year, the discourse has largely focused on its potential use for cheating, automation of simple tasks, and concerns for the future of academic integrity. While there is global disagreement about its potential and potential harms, many educational technology researchers have explored more pragmatic applications, and the opportunity to reflect on wider systemic issues that are a more serious threat to the landscape than an AI tool. From reactionary outright bans to subversive methods to incorporate AI into assessment, it’s important to recognise that similar tools and advancements are not new, and automation in teaching and learning should be examined with a wider lens. So much of the ed tech discourse has focused on ChatGPT for the last number of months, and it’s seeming impossible to escape it.
In their response to the recent advancements in AI with ChatGPT, Williamson, Macgilchrist, and Potter outline the current context:
More generally, they [LLMs] illustrate the extent to which the education sector is currently being washed by waves of developments in AI, datafication, machine learning and automation, which are already exerting concrete effects, raising ethical conundrums and catalyzing acute controversies.
(Williamson, Macgilchrist. and Potter, 2023)
The true complexity of the current landscape goes beyond the current headlines and calls for ethical reflections on the wider issue of ratification and automation in higher education. In their 2021 UNESCO report, Facer and Selwyn advocate for a future of digital education approached with “non-stupid” optimism as they critically reflect on ed tech innovations throughout the years (Facer and Selwyn, 2021). The recent advances in AI require a similar, measured critical reflection. ChatGPT is not in itself a techno-fix, nor is it useful to respond to its potential use for cheating with more AI. Rather, it seems an important time to both consider the ethics underpinning these tools (such as the exploitation of Kenyan labour to remove harmful content) to the potential for improvements to accessibility features for students with the current AI advancements.
In the above exchange, Tim Fawns and Laura Gibbs make valuable points about the knee-jerk response to ChatGPT, urging educators to focus on AI in a wider context and what potential it can have for teachers and students. The types of technological enhancement that Gibbs outlines above are necessary for student agency, and would require not only additional labour for teachers, most of which would either be unachievable technically or because of time constraints. Helen Nicholson from JISC outlines some of the recent advancements in auto-captioning in this post, including some of the advancements in OpenAI’s less-talked about tool, Whisper. Facer and Selwyn stress the importance of optimism in their report, and automation advancements such as this provide a brief moment of respite.
DALL-E, the AI image generator, requires a cautionary lens, especially in relation to spreading potential biases. This short screencast provides some initial reflections:
Moving on to explore for-profit models of automation, Riep’s paper on the Bridge International School model examines the more entangled underpinnings of techno-solutionism and the failures of the ‘out of the box’ model of ed tech in a context that cannot possibly support it. While on a surface level, it’s another seemingly heartwarming story about educational technology reaching those in need, with Riep’s research it is evident not only how the agency of teachers and learners within the model is limited, but also how the complicated rentiership-adjacent model that it’s built on benefits corporations more than the teachers or learners themselves:
Commercial linkages between Pearson, Bill Gates (Microsoft), Nook Media, and Bridge represent an assemblage of socio-material relations, which piece together capital investment and market tools and technologies (i.e., tablet e-readers) to strategically design new education markets.
(Riep, 2017)
Simply put, there is always a need to follow the finances to untangle things, and in this case, the design and structure of the model do not seem intent on supporting student or teacher agency but are more concerned with the solutionism offered by the model.
In Facer and Selwyn’s report, they reflect on the initial flop of Seymour Papert’s LOGO, but the subsequent success of programmes like SCRATCH and constructivist models like the Maker movement and Minecraft (Facer and Selwyn, 2021). The cautious optimism outlined in their report is a lens worth adopting in our current context. While it’s likely that much of the panic around automation will subside, advancements in terms of productivity and accessibility will occur, and hopefully, the current tools and practices will improve both technically, ethically, and pedagogically over time. While many of the current automation advancements are seen as problematic, it’s important to remain optimistic. Consider how a tool like ChatGPT could be improved upon to ultimately have the same impact on writing as SCRATCH has had on coding?
This is a worthy challenge for both teachers and developers alike. How can LLMs be used to support writing? Can it teach students the basics? Can we ensure that it does no harm? There is scope to support students in that writing process, and our role as critics will be to ensure that the future versions and adaptations do their best to address disadvantages, support students, and reproduce no disadvantages:
Directing technology use in education towards recognising and addressing issues of equity, diversity and overcoming disadvantage, as well as the potential for these tools to themselves produce and reproduce such disadvantages.
(Facer and Selwyn, 2021)
Automation in teaching and learning has long been a cause for concern among educators, especially when promises fell short, or proved too good to be true, and recent innovations in AI have elicited a pile-on of techn0-solutionism to tech-created problems. Facer and Selwyn envisage a new approach, “where the tools will have their value precisely in their mobilisation and practice, not despite them”. (Facer and Selwyn, 2021)
References:
Cheney, M. (2023) Academic integrity?, Finite Eyes. Available at: https://finiteeyes.net/technology/academic-integrity/ (Accessed: February 18, 2023).
Facer, K. & Selwyn, N. 2021, Digital Technology and the Futures of Education: Towards ‘Non-Stupid’ Optimism. Unesco, Paris France. <https://unesdoc.unesco.org/ark:/48223/pf0000377071>
Gibbs, L. (2023) [Twitter] 29 January. Available at: https://twitter.com/OnlineCrsLady/status/1619803902983143425?s=20″> (Accessed: 19 February 2023).
Molloy, K. (2023) Experiments with DALL-E. Available at: https://www.youtube.com/watch?v=BxVVHT3oQKI&feature=youtu.be (Accessed: 19 February 2023).
Nicholson, H. (1969) AI & Accessibility: The present and potential future of automated transcriptions, Media and Learning. Available at: https://media-and-learning.eu/type/featured-articles/ai-accessibility-the-present-and-potential-future-of-automated-transcriptions/ (Accessed: February 18, 2023).
Perrigo, B. (2023) OpenAI used Kenyan workers on less than $2 per hour: Exclusive, Time. Time. Available at: https://time.com/6247678/openai-chatgpt-kenya-workers/ (Accessed: February 18, 2023).
Riep, C. (2017) Making markets for low-cost schooling: the devices and investments behind Bridge International Academies, Globalisation, Societies and Education, 15:3, 352-366, DOI: 10.1080/14767724.2017.1330139
Sawahel, W. (2023) Embrace it or reject it? Academics disagree about ChatGP, University World News. [online] Available at: https://www.universityworldnews.com/post.php?story=20230207160059558 (Accessed: 18 Feb. 2023).
Williamson, B., Macgilchrist, F and Potter, J. (2023) Re-examining AI, automation and datafication in education, Learning, Media and Technology, 48:1, 1-5, DOI: 10.1080/17439884.2023.2167830
As always Kate, a very engaging and enjoyable post. I loved some of the multimodal elements such as the video on DALL-E and the inclusion of the tweet about ChatGPT. It’s not just that you have included multimodal elements, however. You use these elements to provoke useful critique and evaluation of the issues.
I’ve been following the discussion about ChatGPT on Twitter the last few weeks and it’s been very interesting. Your video about DALL-E and your audio commentary were also really interesting – thanks so much for that. You are so right – what is in those algorithms? It’s really disturbing, particularly the college students and the point you made about what data you are giving the software to ‘train itself’. Be interesting to see how DALL-E changes, or if it changes as more people use it.
Some questions which your post brought up for me – should technology lead pedagogy or vice versa or is that question itself missing the point, in the light of sociomaterialist perspectives? Is there a genuine ‘need’ in education for ChatGPT or are the developers trying to convince us that there is? In your discussion of how it might be used to improve student writing I couldn’t help thinking how lovely it would be to think that the developers started with issues in student writing and designed their AI to help and support that, rather than how it might make money or be used to access new markets and new forms of data!!
I’d love to hear more about what you think regarding the impact does automation on the role of the teacher. How are structure and agency being reworked in teaching contexts which include automation and AI? Forgive me if you have discussed this in your next post.
Some great hyperlinks to relevant supporting articles – thanks for those.
Thanks for your comments, Noreen. I just opened this post on my iMac (had been working on this on a laptop) and I just realised how much more horrifying the DALL-E images of people are on the big screen. Eek! I would have had much more to talk about there if I’d seen them that clearly.
I agree with your point about a genuine need for some of these innovations. I’d love to see a supportive writing tool developed that starts with those values in mind rather than being money-driven. In my field, I always feel like we’re trying to retrofit or repurpose tools and tech to suit our values and needs. We do this with tools like Turnitin. We take the line, “sure, it’s kind of bad, but it has great feedback tools built in, so let’s use it!” We often have to negotiate those tensions to suit our needs and our status within the university. A few years ago, I coined my take on this as ‘digital pragmatism’. Many of us aren’t in a position to ask those bigger questions about tech in our daily lives or avoid/reject certain tools where policies and procedures are in place. I actually just finished a chapter on this area with my collaborator Clare Thomson. I’m probably not explaining this very well, but will share the chapter when I can. Sadly, it’s a much larger problem in ed tech in general, so many new markets and forms of data as you say coming from a not-so-wholesome place.
I’ve be very interested to see your work on this – yes, do forward it when completed.