AI: The Third Wheel for Artists and Their Creations
How AI Changes the Relationship Between Creators and Their Work
In previous articles I have talked about how originality in itself is not necessarily necessary to produce of work. Much of the great art that has been created over the years have revolved around a narrow spectrum of themes from the Bible and classical works. I also touched upon the idea that the traditional tools of an artist have been independent of outside forces. How an artist masters and uses their tools has been totally within the confines of their own talents and abilities.
The difference with using AI tools is that the tool itself, the platform, is ingrained with the biases, both intended and unintended, of the creators of the platforms. This calls for a new relationship to be determined between the artist and their AI tools of choice.
The very first question to be asked is if prompt engineering is an art form in itself or a craft. AI is really nothing more than an ultra massive prediction machine. The essential skill for the user is to reverse engineer the patterns of operation to produce a useful output by the means of appropriately worded prompts.
It is a technical skill that requires understanding of both the capabilities and limitations of AI systems. While that process may require some careful and artful tinkering to learn and master, it is only to produce results that can be usefully employed in some other more useful or entertaining purpose and not as an end in themselves.
Skilful prompt engineering can help extract better results from AI, but the question always has to be asked: ‘To what end?’ Like quills, keyboards, editing systems, and all types musical instruments, AI is there to be able you to do something else whether it be functional or artistic.
The tools themselves may be beautiful but their existence is predicated on their intended use.
The big difference between other tools and AI is the presence of third party biases of the programmers and creators of the platform. That these biases exist is not in contention and are a major issue in the field. Emily M. Bender’s paper, "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” is a good place to start.
While before there was just the artist and their tool, there is now a new partner in the relationship - the AI programmer. As with any relationship there is always baggage that comes along.
If all the biases in the AI system were there to be seen they could be adapted to and taken into account when engaging on a creative piece of work. We could spot the cultural biases and respond to the necessary safeguards appropriately.
But the problem with biases is that our knowledge alone of them is to create, as Daniel Kahneman wrote in Chapter 19 of Thinking, Fast and Slow an ‘illusion of understanding.’ In short, just because we know about our biases doesn't mean we won’t still act them out. We can’t help ourselves as the heuristics which fuel our biases are so useful in reducing the cognitive load on us as we lead our lives. They are indispensable and we constantly deploy them without thinking or reflection.
Those are the biases we know about and there are plenty of others to which we remain oblivious. The problem compounds with AI platforms. In dealing with what is pretty much an opaque system for the ordinary citizen we have to assume that there are probably a multitude of other indeterminable biases influencing our results.
Prompt engineering, no matter how sophisticated, is simply about mastering a tool rather than creating art itself. Creativity lies in what we choose to do with these tools - in the vision we bring to them and the purposes we put them to. But we have to be aware of the capability for biases inherent in AI systems and their ability to warp reality and mislead us.
With AI as our co-creator we will have to work harder to ensure we have the proper information; otherwise, how can we really be sure about the truth of our work as anything else is just lies, propaganda, and advertising?