Opinion
When essential needs, such as food and shelter, are met, humans yearn next for … stories. Humans need stories to feed on, and we like to tell stories. This sits deep in our DNA.
Homo sapiens are not consistently wise, but we are all interested in news. Different types of news perhaps, as some people like politics, others science, others sport, some the latest high literature, art and films, others celebrity gossip, and many prioritise updates from friends and family, but one thing is sure, and it is evident in our addiction to our phones, we love to be in contact, and we love conversations.
Life can be viewed as a series of successive conversations. Sometimes we are building friendships, sometimes we are persuading, sometimes we are learning, sometimes we are being tested.
The arrival of chatbots, that can pass the Turing Test (that is, they cannot be distinguished from real humans), has dropped a grenade into our world.
I worry about the impact of chatbots on education. I love AI, when it concerns the Nobel Prize winning AlphaFold and the ability to determine protein structures and design therapeutic inhibitors. But I don’t like it when I mark student work written by chatbots.
Many people have talked about cheating and AI but less has been said about the academic despair involved in grading and providing feedback on student work generated in part or exclusively by chatbots.
Marking has always been hard work but sometimes it is very rewarding. But now if I suspect work has been written by a chatbot, or even if it is openly acknowledged that AI was only used for ‘brainstorming, smoothing, and shortening’ (i.e. for everything) I am not outraged, I am just saddened, and ask myself – what is the point?
I still get some satisfaction from marking supervised exams. At least I know I am reading a real person’s story. If the work is good, I rejoice. If it is weak in parts, I feel I can help by providing targeted feedback. In contrast, I have no motivation for providing feedback on chatbot generated assignments. Should I get my own chatbot to do that, and short-circuit the entire educational experience?
The topic of ‘invigilated exams’ is a divisive one. Some of my colleagues are for, and others against. I’m for – I believe exams have shown their value for hundreds of years, I believe they can incentivise students to learn the whole syllabus, rather than just focus on part of it for an assignment, I believe they are harder to cheat on than take home assignments, and I believe their efficiency means we save resources to apply to other important university activities. I know exams are not perfect, but nothing is.
Some people argue for vivas, but with short vivas there is too much luck, as each student only gets a small number of questions and can be thrown if the first one is too hard for them. There can also be bias in vivas, unless a panel of markers is involved. Thus, panel-based, lengthy, vivas can work, but they drain our resources.
I know that in each discipline academics are experimenting with new chatbot-proof assessment strategies that can assure learning. There may also soon be online identity management strategies, vocabulary or keystroke fingerprints, or other technological solutions. So, I’m not wedded to exams, I just think they are the best available ‘at scale’ solution we have at the moment. I’ll change my position on exams when I hear a better strategy.
I also believe that exams are more authentic than many people give them credit for. They are authentic because they enable a conversation. Students and teachers can, at best, be like pen pals who write to each other. Corresponding is hugely authentic because this is at the core of what it is to be human. Humans like hearing stories but they also like writing them, this nourishes us, except when a chatbot intrudes creating suspicion, inauthenticity, and undermining our relationships and the purpose of education, thus diminishing the care that good teachers have for their students’ learning.
Everyday we all exchange stories with our peers. We do it instantly in real time, without using chatbots. We talk to our friends, our bosses, our colleagues, our staff, our students, our parents, our partners, our children. “All the world’s a stage”. We are all actors. But relying on chatbots for this conversation is inauthentic and always will be (though relying on AlphaFold to determine protein structures is fine, as is using chatbots for many other things).
Too much of the debate about AI in education has been confused because AI in general is not the problem, chatbots masquerading as humans is the problem. I want my students to be able to use AlphaFold, and other AI tools, just as they use calculators and various computer programs, but I do not want chatbots interfering in the professional relationship between students and teachers.
If unchecked chatbots may poison human educational connections, and undermine the motivation of both students and staff, so we should be very cautious as we inch forward through ‘tomorrow, tomorrow, and tomorrow’ and the age of AI unfolds around us.
Professor Merlin Crossley is DVC (Academic Quality) at UNSW