Reader, in this column I am going to have a think about artificial intelligence (AI), and its relationship to creativity and the architectural discipline. My reason for discussing it is the recent developments in the presence of AI text-to-image programmes, such as Midjourney, in the portfolios of architectural students, both potential and current. The images in question that have invited scrutiny from me and various peers seem to be a bit… uncanny: the way I imagine an intelligence that isn’t quite human would put things together. It is as if objects are sutured together and then Photoshopped to take out the stitches.

Plagiarism is a key concern raised when speaking to colleagues and peers; how do we tell whether or not these images have been created by the author in question? Interestingly, often the debate is further heated when someone feels as though a student has submitted work that is better than their skill set. At times, the discussions can escalate into moral outrage: how dare they try to pass off work that is better than they have the skill set to execute!

I am always a bit on the fence when it starts to swing that way, and not because I am morally bankrupt, although that may well be true. I think plagiarism is unethical and should be neither encouraged nor ignored. I think my hesitation to see it deploying AI to produce work that is better than a person can do as unequivocally beyond the pale is because, for me, I guess, I have always been in the category of tech optimist. Emphasise that AI can aid problem-solving in the design process and the way in which we look at these images might start to take on new meaning. De-emphasise the meritocracy of image-making and reprioritise the creative process as a journey that is unending. Perhaps, the threat of AI as the tool of the cheat is rightly balanced by other positive (and, of course, less positive) possibilities. This, for me, is the more interesting and valuable discussion: what are the ethical parts in introducing such tools into the creative process?

As I prepared for writing this, I had a look at Midjourney’s website and some of the discussion around its use across creative industries. Immediately, I found examples of its use that have underlined counterpoints to my argument above that are more than valid. Last year, a product design manager from San Francisco had a conversation with an AI-powered ChatGPT about a girl named Alice.¹ Using that dialogue, he developed a storyline about how Alice builds her own robot that becomes selfaware. To illustrate the story, he used Midjourney. The whole process reportedly took a few days and he did not have to draw a single thing. Artists working in illustrating pointed out that artificial intelligence, or machine learning, requires their own work to learn how to make images. Any artist who uploads their work to the internet (and, who doesn’t put their work online?) would unwittingly feed the algorithm.

It is not hard to see how damaging this could be for creatives and their livelihoods. One cannot underestimate this as an issue to insure against. Certainly, when I found myself using Midjourney, researching in the field, the image I created using the prompt “equitybased urban cities and ethical architecture” felt anything but ethical. As someone who hails from the X Generation/very cusp of Millennial, I am not sure I even understand the mindset of today’s architecture school applicant and student. They held iPads as children; do they distinguish that threshold between the body and the machine as I did? So seamless is their skin fusing with the screens of devices while I, when first using the internet, would hold a mouse and navigate it as though I was trying to get a toy plane to fly, and would type as if selecting a floor in an elevator. Once in, I would clump around search engines with little finesse or awareness. Most times, I looked no further than what I first encountered. All those movements had to be learned, all while hearing the dial-up ‘bing’ of yore.

What I think I am contemplating here, with many a segue, is: do I understand the stakes of what digital tools offer architecture? Am I, with my diminishing time left on this earth, even a major stakeholder in an architectural future contending with AI design? How, then, do I conceptualise AI and creativity for this developing workforce, inheriting the problems of multiple crises converging on one another?

When I consider whether or not, generationally, I am equipped to understand the digital native’s outlook on the role of computing in solving design-based challenges, I also wonder if I can truly key into the future of a young person wanting to practise architecture, for whom the climate crisis is at the front door. What sort of tools does such a practitioner need? Often, while I write these columns, I end up having discussions with myself. This is as murky as it sounds. I go over my notes for this text and volley back and forth with questions for myself about what is important. 

I find myself drawn to the section titled ‘Code of Conduct’ on Midjourney’s website.² Under the heading are three bullet points: “Don’t be a jerk. Don’t use our tools to make images that could inflame, upset, or cause drama. That includes gore and adult content. Be respectful to other people and the team.” Apart from a few other lines directing users to content guidelines and some other facts, those directions sum up the Midjourney approach around expectations of behaviour. It is not too far from what most children tacitly agree to on a playground. And perhaps this is the point; the bots are here and those who are making them are, again, leaving the tool prone to weaponising. We have to ask questions but, also, we have to ask whether or not those questions are the right ones.

References

¹ Midjourney, docs.midjourney. com (accessed 31 January 2023).
² Midjourney. ‘Quick Start’. docs.midjourney. com/docs/quickstart.