Responsible GenAI for responsible professionals

Recently I made my first generative AI video, in my own time and at my own expense (you will appreciate why I am emphasising this shortly), relating to a new Data Maturity Framework for universities and colleges from my day job. Please do take a look!

One act play - teaser anon - YouTube

In the course of doing this, three things have become clear to me (in a personal capacity).

First, it is ironic that, as brilliantly as members of our academic research communities strive to develop powerful GenAI techniques that allow such videos to be created via natural language prompts, members of professional services of the same organisations are working with similar ingenuity to block the resulting tools being used for professional purposes, for reasons including uncertainty over copyright, and ethics. I guess this could be viewed as part of the 'creative tension' which helps make our universities the powerhouses that they are!

Second, I realise I am getting old, through a sense of déjà vu... So while I was not around to experience the resistance when Oxford and Cambridge Universities moved to stop teaching the curriculum through the medium of Latin, I can remember, in the 1980s, serious concerns expressed over the ethics of using word processors, and especially spelling and grammar checkers, for students’ academic work; and in the 1990s some equally serious angsting over whether one might get sued for breach of copyright for using a Web search engine, or even whether one required formal written permission up-front to link to a third party's Web site, to avoid that same fate. As drole as it may be to look back 20 to 30 years or more in this way, to me these are accurate historical analogues of the current concerns over GenAI use. I believe precisely no one in the past was sued for using such tools in their work, and that might be why Microsoft is now quick and comfortable in offering its Customer Copyright Commitment, to pay for any adverse judgments, should its customers be sued for copyright infringement arising from fair use of its AI suite.

Third, it is a cliché, but this area is moving incredibly fast. It is almost 20 years since Second Life burst on to the scene, with many universities and colleges investing significantly, buying virtual plots and creating virtual buildings and campuses (in a part of the virtual world that turned out to be adjacent to the ‘adult region’, apparently…). Second Life graphics and visualisations now look faintly comical; but equivalent GenAI step changes are happening in months not years: Midjourney still images, all the rage just six months ago, are already looking a little dated, having been quickly superseded by DALL.E 3; and a similar rapid shift looks on the cards when e.g. the new Sora video engine is released. Who knows what is to follow.

As with previous innovations that have impacted academic and other spheres, whether it be teaching via English rather than Latin, using word processors, or the World Wide Web, there is no going back with generative AI. So, while people's bias, ethical and intellectual property concerns are genuine and of huge importance, addressing them by erecting barriers and issuing commands to stem the tide is as futile now as it was for King Canute. Instead, we must surely be pragmatic, we must embrace and encourage valid use of GenAI by responsible human professionals employing their human judgement, and with a real-world approach to the balance of legal risk.

We should be clear: if we keep looking to block reasonable use of GenAI by trustworthy professionals (like me!), we risk surrendering, without contest, the use of these powerful tools to loony conspiracists on social media, and, far worse, to malign foreign powers looking to manipulate our forthcoming elections. Surely, however well intentioned, that cannot be right for any organisation seeking to act for public good.