Are you satisfied with average?
The age of Generative AI drives us to a crucial question we all need to answer, and this is a question that AI itself cannot answer for us. Should we be satisfied with mediocrity, or strive for something better?
Fundamentally, the Large Language Models (LLMs) upon which tools like ChatGPT and its ilk are based are entirely derivative. They slurp up data from the widest possible sources, shape it into a remarkable predictive engine, which can then return back to you quite astounding results. That is to say that, at first glance, they appear remarkable. Contrary to initial appearances, I argue that the very nature of how these predictive engines are built is entirely designed to respond with a mean reduction of the information it has syphoned up.
ChatGPT will give you the average answer to your query from all the information it has absorbed. On the one hand, this is indeed remarkable as a product of complex mathematics and statistics. However, on the other hand, this could also be described as mediocrity incarnate. To be sure, depending on your query, you may be able to drive your results towards some interesting corners, but they will still simply be the statistical correlates of your query terms. There is no creativity or “out of the box” thinking here.
If one asks GenAI to write a poem or a book, it will do an amazing job at producing the best derivative output it can muster on the topic, but it will never have the capacity to independently arrive at works of the unique quality of Shakespeare, Keats or Tolstoy. It will never pop out a grand work of literature such as Dickens or any other great author of your choosing. It may be able to imitate their style. After all, it has soaked up their work, so this is entirely within its ken. But it can never produce work with a stroke of originality. It will be unable to arrive unaided at the weird world and language of a Joyce or the powerful narrative of a Conrad. It does not have this power. It is an imitator of the highest order.
Now, some may argue that all art is derivative in some way. Indeed, the true greats of their fields often found a way to build on existing works. An important distinction, however, was their ability to subvert, redirect and veer off the beaten path in an unpredictable way. This allowed them to reshape literature, art and even scientific endeavour in new and never before seen ways. Picasso’s early work, for example, demonstrated his exceptional skill in traditional painting styles, showing a mastery of classical painting techniques. However, it was his departure from these traditional methods that marked his contribution to modern art, particularly with the development of Cubism. That kind of subversive and sometimes rebellious originality is not within the capabilities of GenAI. These algorithms simply cannot do this. They can take what exists and predict the best approximation of what is being asked for, but it will always be an average response.
For example, I regularly present at conferences. When LLMs first became the rage, I experimented with asking ChatGPT to assist me with ideas for preparing a speech on some of the topics I cover. The results were, in my view, underwhelming. ChatGPT would return back the outline of a presentation that was quite surprising in its coverage of the topic. But, this suggested presentation—its content and structure—were as average and underwhelming as one could imagine. This was the recipe for boredom, multitasking by the audience and predictability. This was never going to be a presentation I would want to give, or even listen to, because I aspire to be better than average. I want to be excellent. I want to surprise, entertain, provoke and stimulate my audience. I want them to listen and laugh. I want them to go away feeling like that presentation was memorable (for the right reasons). Average will not do for me.
Is using these tools a waste of time? No, I don’t think so. But I do think we need to keep in mind that they are simply tools. They’re good for certain types of tasks and need to be used thoughtfully and judiciously. To use the old aphorism, not everything is a nail. If I’m after a summary of information about a particular topic, ChatGPT will do a fairly good job of giving me the standard treatment in a very thorough and conversational way. I could Google up a storm and collect the information myself and synthesise it into a useful summary, but LLMs are very good at this task.These tools may provide you with a research platform, a starting point, but, when it comes to the creative business, the playing around with ideas and new ways of looking at things and packaging them up in a contextually sensitive and engaging way, ChatGPT will not do. Unless I’m happy with mediocrity on that front, and I am not.
There are plenty of studies, scientific and otherwise, which purport enormous productivity gains through the use of GenAI tools. Again, in one sense this may be true. The speed at which you can produce outputs is enhanced. But, as I’ve tried to argue here, that does not address quality. If your definition of productivity is entirely one-dimensional and focused on time, then you will be highly productive at being average. If you are looking to be excellent, then these tools will definitely help with some of that task, but perhaps not nearly as much as you might hope.
The human mind is still needed. So, as these tools continue to gather hype, we need to make a decision that only a human can make. Are we happy to be mediocre or do we want to be above average, even excellent? No doubt, in some scenarios, good enough is good enough, but in many others this is a recipe for being a boring clone. That’s not what I aspire to be. This may sound a bit grandiose, but I suspect it’s what many of us think when we’re given the opportunity to create and make something. We want our creation to be great. To be worthy. To be of note and something which reflects well on us. We do not want to be boring, merely average, or a facsimile of a thousand other examples.
So, my exhortation is not to let an over-reliance on AI or a subjugation of your creative capabilities to the great mediocrity that is baked into AI cause you to lower yourself to something less than you can be. Be anything but average.