By Ross Jordan and ChatGPT
This year’s blog (one a year is quite enough from me) is co-authored for a change. My co-creator is ChatGPT (https://chat.openai.com/auth/login).
If you have not heard of ChatGPT yet, you surely will sooner rather than later. This is an open-source (for now) Artificial Intelligence tool known as a Generative Pre-Trained Transformer. It was released in November 2022 and secured 1 million users within a week (a feat that took Facebook 10 months). It is by no means the only tool of its type but it is the one receiving the most widespread press. I suspect many of my academic colleagues would rather I did not add to this press, but my sense is we cannot take the resistance approach exemplified by the New York City Education Department (https://www.nbcnews.com/tech/tech-news/new-york-city-public-schools-ban-chatgpt-devices-networks-rcna64446).
The first question I asked my Co-Author was; “What impact might Artificial Intelligence have on University assessments”. The next section of the blog is the response generated in a matter of seconds.
ChatGPT: “Artificial intelligence (AI) has the potential to significantly impact university assessment strategies by automating certain tasks, such as grading written assignments or administering exams. This could allow professors to spend more time providing personalized feedback to students and less time on administrative tasks.
AI could also be used to analyse student performance data and provide insights that could inform how professors design and structure their courses. For example, an AI system might be able to identify patterns in student performance that could indicate the need for additional support or resources.
However, it's important to consider the limitations of AI and the potential risks of relying too heavily on it for assessment. There is a risk that AI systems could perpetuate existing biases or make mistakes that could have negative consequences for students. It will be important for universities to carefully consider the ethical implications of using AI in assessment and to ensure that it is used in a way that is fair and transparent.”
I was giddy with this response and insisted my partner experience the same excitement by asking ChatGPT a question. She was convinced that her question could not be addressed well, but in a similar timeframe, the GPT did indeed produce a lesson plan for a 30-minute standing-Pilates session for older people.
How is it doing this? Basically, it is self-learning and accessing a lot of internet data (approximately 300bn words of it), and at the risk of offending AI experts (of which I am not), it is looking for patterns. In a similar way all humans are engaged in pareidolia. Seeking to make sense of the world from the clues they interact with based on experience, learning, and a whole set of complex emotions and cognition (the last two being trickier for AI, but definitely on the radar). If you would like a fuller explanation including how things have leapt forward recently here is a good resource:
Therefore, this is a bit like when an academic is marking a student’s work and is looking for patterns to reassure them that the student has understood the context, frameworks and theories they have applied, supported their thinking with reliable sources, met the marking criteria, and in many subjects displayed originality of thought and criticality. This is a skill, which academics develop over time, along with the means to give constructive and supportive feedback (feedforward) to help students improve. However, the academic workload is cited by many as having increased to be burdensome over the last few years. Students also desire for more detailed and fuller feedback for all sorts of reasons including the oft-mentioned consumerisation of university study, and a focus on employability in a challenging job market. Therefore, my interest was to see if ChatGPT could help provide elements of feedback with a consistency and focus, which time resources often work against, thus buying back time for richer focused human feedback.
For clarity, I haven’t tried this yet and continue to read and provide feedback in full in the traditional manner. Why?
It feels like a Genesis of the Daleks moment (apologies for the reference for any reader under the age of about 50), where the Doctor goes back in time and has the opportunity to kill off the Daleks at birth by touching two wires together (do go and seek it out as it is much better than all the new Dr. Who stuff – another sign of my aging; https://tardis.fandom.com/wiki/Genesis_of_the_Daleks_(TV_story) )
The Doctor decided he could not do it, as it will rupture the space-time continuum (of course), i.e. it will affect everything to come and change history.
However, in my view, ChatGPT and its compatriots will change the future in many spheres including academia regardless. Many others will be prepared, and almost certainly have already, touched the wires together. Because, of course, the next step is for students to ask GPT to write their essay answers. This is the nuclear option in terms of the weaponization of plagiarism. Elon Musk’s response, in contrast to the New York Education Dept., was to tweet; “no more homework”.
As a test, I thought I would ask ChatGPT to address one of my own (very simple) assignment briefs. I input the adapted instructions (for ease of machine understanding rather than human understanding), and the marking rubric for an A* grade, and pressed enter. The assignment requires students to generate a new business idea and pitch it, concisely, in just 250 words. A good answer shows originality, research, acknowledgment of competition, and the scope of the opportunity. My Co-Author offered the following:
“My business idea is a subscription-based meal delivery service that focuses on delivering fresh, healthy meals to busy individuals and families. The key features of our service include fully customizable meal plans, with the ability to choose from a variety of ingredients and options to suit specific dietary needs and preferences. We also offer the convenience of home delivery, as well as the option to pick up meals at designated locations.
Our target market is busy professionals and families who prioritize their health, but may not have the time or energy to meal prep on a daily basis. Our value proposition is that we make it easy for our customers to eat healthy, without the hassle of grocery shopping and cooking.
We differentiate ourselves from competitors by offering a wider range of ingredients and options, as well as the ability to fully customize meal plans. We also prioritize the use of fresh, high-quality ingredients and sustainable packaging.
In terms of potential value generation, we see a clear demand for this type of service, as more and more people are looking for convenient and healthy meal options. Additionally, by offering a subscription-based model, we can generate recurring revenue and build a loyal customer base.
Overall, we believe our meal delivery service has the potential to be a successful and valuable business venture.”
Reassuringly this is not an A* answer, in fact, it is barely a pass, but it is an impressive and well-expressed start nonetheless. In the same way that students may type their assignment question straight into Google to get started (no really – do they do that?) this might be a new jumping-off point for many. Re-entering, or using GPT terms, regenerating the answer from the previously produced output with additional instructions each time (such as identifying the principal competitors to this business and quantifying their market share) improves the response each time. This particular assignment is a tough one for AI, at present. It requires, imho, the in-class experience of the taught environment to support students originality of thought and evaluation of opportunities. In other words, it is an assessment strategy designed to work with an enterprise pedagogy, which is now a typical approach of an academic course focused on start-up businesses. However, the notion of an enterprise pedagogy continues to gain traction across many subject areas, sometimes to the nervousness of disciplinary specialist who incorrectly tie it to the start-up notion. Whereas an enterprise pedagogy is really about encouraging an enterprising mindset (innovation, appreciation of change, risk awareness, resilience, and understanding of the self-etc.) and is closely tied to more accepted and ling standing terms such as applied, experiential, problem based or project-based learning, amongst others.
This is the sort of learning which happens most readily, in my opinion, in the real world physical space (didn’t want to say classroom as that conjures a conventional image). So this is my key takeaway from this brief dalliance with ChatGPT. It is a further incentive, if one were needed, to revisit our approaches to assessment as academics, in an environment where concerns around student attendance and/or engagement are greater than ever. An experiential enterprise pedagogy is a dish best served on campus, and perhaps once the crisis and fight around AI’s impact on assessment subside we might even see it as an aid to an enterprise pedagogy.
As a footnote to the last sentence, I would like to share a few links (as always) that may take you further if your interest has been stimulated.
The other reason AI is of interest to me is because of my involvement with start-ups and the investment community. It seems clear to me that we are at an infection point, of impact and thus investment in AI (a proper one, not the VR glasses nonsense of the past). Arguably, this is the trigger of the 6th wave of disruption mentioned in my blog pre-Covid.
Take a look at one of my trusted sources to see some exciting start-ups and funds in this area:
What I find particularly interesting here (and given the nature of the test assignment mentioned earlier) is the potential for AI in a creative sense, in terms of images, design, and voice generation). This is potentially creative in and of itself, but at least opens up the potential to be creative for others in many different ways. You could for example ask DALL-E, an image generator version of ChatGPT to produce a logo for your new business or the image in the header of this blog (which I did):
A means by which this technology could help all businesses is in accelerating the integration of existing knowledge data within businesses, in the interest of producing bespoke process solutions more cost-effectively. In other words, and to complete the circle of this blog, accelerating learning. Take a look at London-based https://qatalog.com/ or Swedish company Sana:
I’ll leave the last word to my co-author in answer to my question, how significant will Artificial Intelligence be in 2023 for educators and new businesses:
“Artificial intelligence (AI) is expected to play a significant role in the education and business sectors in 2023.
In education, AI can be used to personalize learning experiences for students, adapting to their unique needs and learning styles. AI-powered tutoring systems can provide immediate feedback and guidance, while AI-powered tools can help educators better understand and address the needs of their students. In addition, AI can be used to analyse large amounts of data, such as student performance data, to identify patterns and trends, which can inform decision-making and help educators improve their teaching practices.
For businesses, AI can be a powerful tool for automating tasks, improving efficiency, and making better decisions. AI-powered customer service chatbots can handle routine inquiries, freeing up human agents to handle more complex tasks. AI can also be used to analyse and interpret data, such as market trends or customer behaviour, to help businesses make informed decisions about their operations and strategy. In addition, AI can be used to optimize business processes and supply chain management, leading to cost savings and improved profitability.
Overall, it is clear that AI will have a significant impact on both the education and business sectors in 2023 and beyond.”