In my Department’s recent meeting about the last batch of BA (or, as we call them, TFG) dissertations we have assessed, we already considered the possibility of suppressing them in the near future. This is not for us to decide, but we work on the assumption that, sooner or later, our school or UAB, if not all universities in Spain, will have to rethink the use of BA dissertations. Perhaps all universities around the world.
As you may imagine, the debate has everything to do with the LLM’s, beginning with ChatGPT, but also other tools that have integrated AI, such as Grammarly, a favourite among our second-language students. A problem we failed to anticipate was posed by the method used by one undergrad student, who drafted the whole dissertation independently, but then used ChatGPT to improve the text, sentence by sentence. This could be acceptable in a situation in which a non-native author needs to make sure a text is fully intelligible to native speakers of English. The problem is that we judge the actual command of English of each student, and, in that sense, using ChatGPT or any AI enhanced editing tool is incorrect. You might argue that Word also integrates tools that automatically correct spelling and grammar, but these have very minor capacities in comparison to what generative AI can do.
The use of AI becomes immediately apparent during the oral presentation, or viva, of the dissertation when students’ English is often seen not to correspond to the much better written TFG. This is why we have proposed two innovations. One is that the mark awarded for the proposal be based 50% on the written text and 50% on the ensuing tutorial with the supervisor. We have a tutorial anyway based on the proposal, so why not use it formally for assessment? The other innovation will consist of using a bigger part of the 30-minute viva for a conversation between the student on the board on the dissertation. ChatGPT appears to be linked to Canva, so that it can automatically generate a PowerPoint (well, the Canva equivalent) and the text accompanying it, which students then memorise. ChatGPT cannot answer questions addressed by the board to the student during the viva, so we’ll ask some more instead of the couple which we usually ask.
Asking students politely to refrain from using LLMs does not work. Threats about failing their dissertation don’t work, either. They have simply internalised that using gen-AI is like using any other tool we authorise, like those embedded in Word. I never tire of explaining that if you’re going to be judged by your possession of certain skills, anything that gives a false impression is cheating. If a third party takes an exam instead of the student, this would undoubtedly amount to cheating. The same applies to elements that function in practise like substitutes capable of doing the full job, or that correct or edit text, applying skills the student lacks. This is also cheating, like being on drugs to boost your athletic performance.
BA/TFG dissertations were introduced in Spain with the European-style degrees in 2008/9. The old style Licenciatura already included a final ‘tesina’, of which the newer TFG is descended. Indeed, many universities still keep the older label. In my times as an undergrad, writing a tesina was not required, and though I can’t find information about their suppression, my guess is that they were gone when the Spanish university became massified in the early 1980s with the access of working-class students. This is, precisely, a factor that was absurdly ignored when the new grades were planned, for it was taken for granted that all departments would have sufficient staff to tutor all the final-year students. As far as I recall, we haven’t hired any external tutors for the only purpose of supervising TFGs, though I know from one of my ex-students that this is a habitual practise in the UK.
The supposition behind the TFG dissertation is that the huge investment of teaching hours it entails is necessary because that TFG validates the skills students have acquired throughout the degree. In the particular case of the Humanities, and specifically of English Literature, these skills can be perfectly mimicked by ChatGPT and other LLMs. This is very unfortunate because there is no other set of skills we can teach, and if we don’t teach those we use our profession will die. To use an analogy outside my field, suppose AI-controlled robots become so skilled at surgery that they surpassed human surgeons in all areas. Suppose next that all schools of Medicine stop training human surgeons to operate on patients. In just one generation, all the expertise accumulated might be lost. So, yes, computers can now write BA, MA or PhD dissertations but if we go down that road, the loss for humanity will be incommensurable.
What should we do then to safeguard intellectual skills? Well, to begin with, we should suppress the requirement that all students write BA dissertations. The obligation to write them is an open invitation for the weaker students to cheat and is resulting not only in a waste of valuable teaching time, but also in a system of student policing ten times worse than the always awkward monitoring of plagiarising. I insist that being a teacher is like being married to a spouse that will inevitably try to cheat, which is not a good basis for any marriage. Let’s keep the TFG dissertation only for the better type of student, who won’t cheat, and let’s introduce a pseudo-honours system. Let’s ask for a B+/A average mark before issuing invitations for students to write a TFG dissertation. There is always the chance that even good students will cheat, but this is less likely, as they’re proud of their achievements. Of course, this possibility can never be entirely ruled out. At least, with fewer undergrads to tutor, we should save precious teaching hours. Finding and encouraging these honours (and honourable) students, besides, is of extreme importance, as they will become the true heirs of the skills we value so much.
Almost two decades after the introduction of the skills-based degrees, we’re beginning to notice that serious mistakes were made. One of the major ones was the cavalier disregard of knowledge and the downplaying of any skills connected to memory and memorising. The main (wrong) assumption then made was that if a student learns to do something, knowledge will follow, which is not necessarily the case. To give an example, students are being taught to identify the main periods and authors of English literature, a key skill in our degree, but there is no requirement for them to study, memorise, and remember them. It’s perfectly possible for a student with good grades in, say, Romantic Literature, which we teach in the second year, not to remember who Byron was two years later, by the end of the degree. Grading is based on textual commentary, applying close reading, and, while this is positive, it’s not always conductive to acquiring information or knowledge. In the tutorials I had last June with my fourth-year students, many had problems to remember which authors and books they had written about only a few months before.
The only way to make sure students don’t use AI, have good communication skills, and possess a sound knowledge of the content we teach them is an oral exam on each of the main areas of the degree (Language/Linguistics, Literature/Culture). In short, I’m proposing that we use the Italian system for a final exam based on the whole degree instead of the TFG dissertation. Among all European Union countries, the Italians have the largest experience of examining students orally; we should perhaps consult with them and see what advantages and disadvantages follow from that method. For the final oral exam students should revise the basics of all subjects taken and train themselves to answer questions about them. As for the required resources, we are already using many teaching hours for the TFG viva that could be used for the oral exam. The time students employ to prepare their TFG should be invested in studying. Honours students should also take my proposed oral exam and write the dissertation as an extra, though I see this needs fine-tuning (you might make TFG dissertations mandatory to take an MA degree, for instance).
I can imagine the resistance to this proposal, but I don’t see how else we can guarantee that the students actually possess the skills we certify in their diploma. I also believe that we need to be more realistic about the skills we teach. We train students as if all of them should eventually write a doctoral dissertation, which is like teaching kids to dance assuming their target must be the Bolshoi. This doesn’t make sense. Our aim should be to make sure that students with an academic inclination are given the help to develop the necessary skills, while we make all of them proficient learners capable of teaching others at secondary-school level, or develop professional careers other than academia.
This might sound old school, but texts cannot be read adequately, much less critically, if information is not retained and knowledge accrued. My view is that in the concern to boost academic skills and avoid memorising ‘useless’ information, we have undermined the methods for students to acquire actual knowledge. With the current TFG, we test them on key academic skills, but, to be honest, I don’t see how those apply to real life, nor what they mean without some basic knowledge. I don’t mean that English Studies graduates should be able to recite the complete list of Dickens’s novels or Shakespeare’s plays, but they should be able to discuss how English Literature has evolved and is currently evolving, rather than write 20 pages commenting on just one text. Even in the best-case scenario, that of the brilliant student who gets an A+ for their TFG, this doesn’t mean that they know much about English Studies.
BA/TFG dissertations are on the way out, that much is clear. In the meantime, we need to think of alternatives that help us to correct the direction taken back in 2008/9. I’ve supervised many brilliant TFG dissertations, and I’m very proud to have helped a number of BA students to publish their work in the journal I co-edit, Hélice. Yet, I also know that less gifted students do not really consolidate their academic skills by writing a TFG and use up, I’m sorry to be so blunt, plenty of teaching time which could be better used. As a second-language degree, we attach an excessive importance to writing, to the detriment of communication skills, which are hardly practised in our crowded classes and are only assessed in the instrumental language subjects and in the TFG viva, for about 30 minutes. Testing students orally in all subjects is just impossible, but it doesn’t make sense to trust the testing of all their academic skills to written language. We could get by doing this until 2022 when ChatGPT erupted to change higher education perhaps for ever. Let’s talk to the Italian academics and start considering how to test our English Studies students on their actual communication skills, and in their own English.