Many students are actively using AI as a personal tutor, and this may be applauded as a commendable use: not asking AI to do the work for them, but to tutor them on how to do the work themselves.
This approach can provide one to one support unable to be replicated in a busy, time-restricted class. Enabling this support is the provision of a wide range of accessible content released to students such as slides, lecture recordings (often with easily available transcripts and audio files) and practice assessment questions with solutions.
To receive AI tutoring, notably with good intentions, students are uploading this content into AI. With little control over this, AI could “learn” from this uploaded content and reproduce content based on it, even (and scarily) using a replica of our voices.
Originality
Aside from effectively “stealing” original content, this has a further future potential implication. A Guardian article recently circulated showing students unhappy being taught with AI generated content, noting they didn’t want to pay for tuition that was comparable with what they could get from Chat GPT. However, if students “feed” AI our generated content, and the AI reproduces similar for others, could we find ourselves being accused of not creating original content when the reality is that AI should be being accused of taking our content?
This is not, however, just about protecting individuals. Universities also need to think carefully about what differentiates themselves from AI, so they remain valuable and attractive to future students. Evidence is showing that an increasing amount of content available on the web is generated by AI, with AI generated articles estimated to have recently surpassed those created by humans. Originality and creativity, along with an effective (human centred) learning environment are likely to be key features which differentiate a university education from one which AI can provide and therefore these features need to be protected rather than becoming AI fodder.
We think comparisons can be drawn to the creative industries where artists, authors and others are seeking protection for their original work being used to train AI models and reproduce similar content without their permission and without compensation. Writing a good lecture, for example, includes creating visual and oral content. The way in which content is explained, examples from personal experience incorporated and the visuals which are created can be original and personal to the lecturer. Should we be considering such content as intellectual property and supporting lecturers to protect their own original work?
Who “owns” a lecture?
We do, however, recognise the issues with this comparison to the creative industries due to additional challenges around originality and ownership of academic material. A lecture, for example, is likely to draw on academic knowledge or theories already documented in a textbook or journal which is a step further than an artist taking inspiration from a previous artists style. Furthermore, the lecture content itself may pose copyright challenges, particularly when available online, which first need to be clearly understood. Adding a final layer of complexity is the rights of publishers and the commercialisation of the underpinning content. Recently, a number of academic publishers, such as Taylor and Francis and Wiley have signed multimillion pound deals which allow AI companies, such as Microsoft, to train their AI models with their content; notably often without the authors consent. This means that even if the lecturer had themselves written the underpinning textbook or journal (likely for no fee!), copyright agreements with their publishing company may negate any protection sought.
Notwithstanding the identified complexities, some initial thoughts on potential considerations for protecting academic content include the following. Firstly, encouraging universities to review and update their regulations to detail appropriate use of lecturer generated materials in relation to AI. Our guess would be that most current guidance is based around appropriate use of AI, how to acknowledge its use and the process for misuse with little specific consideration for protection of teaching content. Once appropriate use is clear, awareness of this with students needs to be communicated so they are not unintentionally training AI models without the creator’s permission. Our experience would indicate this is not even considered as a potential issue by students.
Perhaps though this does not go far enough – and realistically it would be very difficult to monitor. It is likely some students would, even if they know they shouldn’t, continue to upload material to AI. Do we therefore need to look at applying protection to materials being used that reduce the risk of this being copied by AI, similar, for example, to watermarks used by artists on images? Or, may we need to take this further and consider and advocate for legal protection against AI for lecturers, again similar to those being pursued by the creative industries.
We are no legal experts, but an initial search, including Jisc guidance, reveals that lecture recordings, for example, are protected under copyright, owned by the employer if created in the course of employment unless there is a contractual agreement to the country. To protect both the individual and their institutions future, should universities therefore be taking action to ideally prevent, but also pursue, any infringements by AI? Of course, the rules are complex and much of the available guidance doesn’t specifically relate this to AI and hence this needs to be updated for the environment we are now working in.
It is important to acknowledge this is a thought piece provoked by experiences and discussions in the classroom. We are not experts on the workings of AI or the legal position around this, but we think it is important to initiate discussion in this area and welcome comments, experiences and feedback to understand opinions, practices and potential solutions further.
Perhaps this is fighting a losing battle, but we may regret not trying.
“Should we be considering such content as intellectual property and supporting lecturers to protect their own original work?” Yes.
What is a lecturer? Once a upon a time, they read out rare books from the university library and added their own commentary, which the students copied ad verbatim. The exposition of a subject or topic, the presentation of an argument or a model answer, the design of an experiment or an artefact is creative work. Lectures were not published and made available for copying. As printing became ever cheaper, publication of lectures was a new source of revenue for universities and academics, but it was a mistake, particularly if the academic could not regenerate their work through research. The codification of lectures for students has become a substitute for their own study and learning, as if like school, all you need to do is to memorise the information as facts without becoming the person you choose to be employed as.
I do not see how you can prevent students uploading original content unless you require students to sign a non-disclosure agreement and are willing to enforce it with litigation. The practice of withholding information from publication already occurs. Students have been habituated by school, textbooks and examinations to expect complete exposition, which they then copy and think is knowledge, as if receiving an impression on their minds (itself a consequence of the university entrance papers). The openness of modern science originated the trend, accelerated by the printing press. Digitisation has made replication much much worse.
The keeping of secrets was once a requirement for membership of the first college at Oxford. Who have you let in?
Good question Jonathan – who have we let in? Everybody it may seem. I think what makes this difficult to accept are the drivers, as you outline, for change are underpinned by economic decisions rather than improving and widening education which would be more palatable…
Lecturers vary considerably in their approach to the provision of learning materials. Some openly share extensive resources across VLEs and in other formats, while others adopt a more guarded stance. There is typically no standardised practice in this area when it comes to how learning materials are presented to students, even if certain expectations may exist at institutional level.
Moreover, lecturers often rely heavily on sources that are not their own, as noted in the article, and this reliance may be particularly high at Levels 4 and 5, where foundational content is delivered. At Levels 6 and 7, by contrast, there may be greater expectation that lecturers contribute more of their own original material. As such, any discussion of protection or ownership may need to recognise a graduated differential in what can realistically be claimed or safeguarded across levels.
Furthermore, where lecturers have already published their original research in journal articles, conference papers, or other formal outlets, they may have already ceded certain rights or control over that material. In such cases, it may be too late to attempt to reclaim control in the manner suggested.
Thank you for your comment Paul – I agree, it is very difficult to entangle original content from other content or even define what is truly original – is for example presenting an existing theory in a new way with new examples of applications original or not? How would this compare to an artist being inspired (but not copying) a previous artist?