Why I don't use AI

Being a designer working within the space of technology and computing, I felt it was necessary to outline the reasons why my projects are made without the involvement of AI. The points mentioned are not exhaustive nor are they in order of their importance.

Learning and consideration is central to my process

Being someone who enjoys research and exploration, learning is a crucial part of what I do. The act of learning, struggling with problems and searching for solutions is what allows me to grow and improve my skills as a designer. I really consider myself to be a learner (for lack of a better word) before anything else. To use these AI – or rather Large Language Model – tools within my process would be robbing myself of the opportunity to learn.¹

These LLM tools are highly successful at removing friction within a workflow, I've seen it used anywhere from the ideation process to finished prototypes in a project. There is much talk of what this gains in terms of speed and efficiency, but I have rarely seen questioned what is lost in the consignment of this work to machines.

The use of LLMs I feel largely ignores the point of the design process in the first place, to consider and solve a problem. LLMs are tools for speed and efficiency, the philosophy surrounding that is to treat everything as an optimisation point, but it's impossible to optimise the design process for either of those without neglecting the act consideration.

If only it was just that

If it was simply pedagogical reasons that prevented me from using LLMs, perhaps it would be far easier to accept. However, the ethical issues that surround the field of AI are severe, making decisions to use this technology a far more complex endeavour.

The environmental impact of the AI industry is the issue most people are likely aware of, and for good reason. We're all well aware at this point of how much energy AI datacenters consume and the pollution that comes from them.² A common counterpoint seems to be the use of local LLM tools that run on your device, but any argument for these seems to conveniently ignore the impact of training the models they are distilled from.³. As we hurtle towards a climate catastrophe, it feels innapropriate to endorse or make use of technology like LLMs that are so transparently in contempt of our planet.

The work of Timnit Gebru is also especially prudent, her research has highlighted many of the ethical social issues of the AI field. Notably she has highlighted the racial and gender bias that exists both within the trained models and the industry as a whole. She has authored a number of papers that show the scale of impact on marginalised groups as well as the root of the industry's ideology within 20th century eugenics.⁴⁵ This has widely been supported by independent reporting that for example describes the impacts of datacenters on marginalised communities and the role African data labourers plays in the training of models.⁶

I find these to be compelling reasons to both condemn the AI industry and reject their offerings. Software like LLM tools are no longer made by a small group of people in a room, they are now part of a vast, sprawling infrastructure that has chosen to actively exploit marginalised groups in order to maintain the appearance of profitability.

In Conclusion

I hope these points bring more context to my practice in and around technology. Equality and sustainability are incredibly important to me and vital to my practice, which is why I won't use AI technology in my work or daily life. I strongly believe that my integrity as a designer is upheld not only by how I consider the direct users, but also those outside of the perspective of a project that will be impacted by the choices I make.

Despite this opposition, I still keep myself well read on what people and companies are doing with LLMS. I hope you will interpret my arguments not as someone burying their head in the sand, but rather as someone who has critically analysed the field of AI and refuses to participate.