“All your base are belong to us.”
That phrase is a meme that has been doing the rounds on the Internet for many years. It refers to a bad — but comical — English translation of an original foreign phrase within a game, whereupon the alien invaders were victorious .. in all but mastering grammar!
Now we face very different threats of alien incursions, and of our own making. I refer to the threat of incursion into human creativity by generative Artificial Intelligence (AI). The invader’s grammar may be meticulous this time, but a well-spoken invader makes the invasion all the more insidious and seductive nonetheless.
I am a writer. A human one at that (I assure you!) And as a writer, the latest buzz in the media about AI systems generating content is of huge interest to me. It leads me to examine what this may mean for the future of my creative domain - one hitherto populated solely by human creatives and their imaginative output. In truth, I don’t consider AIs to be classed as creatives at all, just sophisticated mimics. Let me explain; but to do that, I first I need to clarify something about AI.
I dislike the term AI used to describe these modern tools like ChatGPT. I much prefer the term ML - Machine Learning. AI is a term we are familiar with from science-fiction and comes with the mental images of robots seeking to wipe out humans. It is an evocative term, laced with emotion, that has fuelled the present hype about these systems. We perceive these writing tools as “intelligent”, as if they are on a par with human intelligence, or - worse - could eventually be more intelligent than human writers. We react emotively to the thoughts of AI writers coming to take our jobs, an automated authorial intelligence, perhaps even an ultimately authoritarian intelligence that will dominate human intelligence itself and tell us what to think and do. This is not an entirely ridiculous belief, but the reality - as always - is more nuanced, but perhaps no less threatening for it.
Tools like ChatGPT are not “intelligent” as we think of our own conscious and imaginative intelligence. Any appearance they have of intelligence is gathered by a process of learning. Computers were hitherto programmed procedurally, where programmers would have to define a complete set of logical steps and behaviours for a computer’s software programme. But as the processing power of computers became greater, consequently the tasks assigned to them more complex. The steps necessary to solve these tasks could no longer be defined and codified as a set of steps by humans. So the leap was made to get the computers to derive the logic to solve a task themselves, based on examples. Take the task of finding objects in an image: how would you define a set of rules to identify, say, a bicycle, in any and every image one appears? We humans can do that quite well, and quickly, and effectively. But if we had to list an exhaustive set of procedural rules to allow a computer to do this, we would quickly come up against the limits of this approach. How do you define a wheel(, and just wheels used on bicycles at that?) Now define how must two of them be positioned relative to each other to constitute a bicycle? Must they be joined by something? What? And what if the image is of a bicycle head-on? Should a bike be defined only by handlebars and one wheel then? You quickly see how difficult it becomes. Wouldn’t it be much easier to give a computer a huge bank of images that we know do and do not contain a bicycle, and give it the answer for each one, and let it derive its own set of rules to determine the presence of a bicycle in future images?
That is Machine Learning. And that is what we are currently faced with. ChatGPT is more sophisticated but is essentially just a vast databank of human writing, codified to understand how groups of words relate and what they mean semantically, so it can generate similar patterns on demand. Would you call that intelligence? We confuse it with such, because it does involve a learning process, which we consider a very natural and human thing. Is it not, for example, the technique used by a child to build up intelligence? Or is the child’s instinctive search for learning — the drive to learn — the true natural intelligence? When does a child’s mimicry cease and become sentient individuality? The philosophical questions ML and AI pose are certainty deep and will be with us for a long time. But for now, let us deal with what we are faced with - content generating AI that mimics human-generated output. And let us examine (intelligently!) what this means for us.
Human development, even human ideologies (for all their manifest flaws) have always depended on the outlier, the free thinker, the rebel. New ideas that didn’t fit the previous mould were proposed and their influence then flavoured the direction of human society. What hope for development if the majority of our consumed reading is derived from the body of what has already been done before? Human development will atrophy when our awareness and understanding becomes limited to the regurgitated uninspired ramblings plucked from some consensual collection of Gestalt thinking from the body of all writings on the Internet. Or the put it simply: will we just get more of the same, all the time. Nothing new or original. Nothing novel to push us forward. It is only the human mind that can spark true, random, chaotic, novelty. That’s not even to address the only human minds at play in this who will be the invisible curators setting the bias and filtering parameters for the mechanisms to effect some kind of social control.
Will we trust anything that is not a face-to-face in-person interaction with an actual person.
The watch word for the coming century will be authenticity, and the desire for it: authentic words; authentic images; authentic news; authentic facts; authentic food; authentic ideas; and authentic thought. Between API generated content, and our modern social penchant for re-writing or cancelling historic writing to appease modern social mores, authentic, physical, historical print will become prized again and even more so. Will authentic human-written words in old books become the vinyl of writing, prized for their quaint, but relatable, realism? While any digital words or modern print be mistrusted for its human credentials. Only time will tell.
Such a desire to reconnect with what is real in human literary history is but a part of a greater movement that I see beginning all around us. We live in a word of not just artificially generated literary content, but also artificial imagery (deep fakes), and artificial news and friends (social media bots). Our love affair with technology is leading many to realise that it is being used to manipulate us, control us, and potentially - with CBDCs and Digital IDs - ultimately to enslave us, and take our freedom, unless we stop the direction we are travelling in.
What I see emerging is a movement, a Neo-Luddism. There are those who want to arrest the inexorable current dragging us to place we would regret going. And it is technologies, like AI, that will be used to pull us there. Oh I don’t expect they will succeed in seizing up the cogs of change any more than the original Luddites managed to stop technological progress, but perhaps they would succeed in starting a conversation and prompting a change of gear in our relationship with technology. They are more of those free-thinking human outliers that have steered the coarse of human development away from the rocks many times in the past.
All we need to do is ensure they be heard, and with their own authentic voice.
That is no easy task. But, as always, it starts with the individual. This writer makes a writer’s pledge, here and now, to never use AI generated content, in part or in full, in his writings. It will always be authentically me. Such as it is. And sometimes with grammar mistakes … just to keep it real, of course.