The written word is too important to risk an AI takeover of it

The written word is too important to risk an AI takeover of it


In his book Guns, Germs and Steel, Jared Diamond explains why the transition from a nomadic hunter-gatherer lifestyle to settled farming was such a major shift in the human story. Once humans learnt to farm, they could put down roots, both literally and metaphorically. Once they were able to grow food instead of having to forage for it, they were forced to organize themselves better, develop specialized tools and skills to use them, and learn to rely on fellow humans. As these agricultural communities grew, people began to differentiate themselves by the skills they had and roles they performed, forming guilds and ‘caste’ groups based on these. This gave rise to the complex modern society that we inhabit.

But it was not until the invention of writing that humankind was able to transform itself into a civilization. While trade-based guilds implemented apprenticeships that were designed to pass skill-sets down from one generation to another, writing allowed information that resided only in the memories of those who shared it to be given permanence, so that anyone with access to this knowledge—from mundane administrative records to epic cultural narratives—could disseminate it much more efficiently.

In her book The Greatest Invention, Silvia Ferrara refers to writing as “… sound made visible and tangible.” It is, she argues, a means to interact with the senses that, while not an innate human faculty, remains essentially human. This use of symbols etched or inked on a surface in order to represent information that until then had only been described through language and actions extended human capabilities in unprecedented ways.

It allowed each individual human experience to enrich the ‘hive mind’ of humanity with recorded facts, experiences and knowledge, enabling the storage of information traditionally passed down from one individual to another, so that it would exist outside the human brain in a form that could be accessed across generations. Writing is the means by which humankind has been able to create an external repository of knowledge that exists independent of the individual, and which eventually led to many of the advancements we have experienced in various fields of human endeavour.

That said, writing was not without its limitations. When information had to be inscribed painstakingly by hand, only those who had access to handwritten manuscripts could benefit from this new invention. It was only after the invention of the printing press that we had a radical transformation of access and witnessed a dramatic increase in the volume and velocity at which knowledge was disseminated. This set off a series of technological advances that successively deepened humanity’s access knowledge—all the way to the internet, which today offers anyone with web access the ability to locate and use the sum total of human understanding in quantities that far exceed the ability of any given human to consume in a lifetime.

In many ways, this rapid acceleration of information availability is the reason for many of the problems we face today. There is a limit to how much information the human brain is capable of processing, despite its remarkable capabilities of analysis and comprehension. Given the vast volumes of raw information that is now available on tap, we struggle to contextualize and synthesize it all in a way that makes sense.

Which is where Large language models (LLMs)—our most recent innovation that uses the invention of writing—comes into play. LLMs are trained on vast amounts of written information scraped off the internet—to the point where they are capable of providing answers to pretty much any question posed to them. Since LLMs are capable of holding in most contexts more information than possible for humans, given the limitations of our brain, they are capable of augmenting our human abilities to process, contextualize and utilise information in ways that human minds simply cannot.

LLMs also internalize the form and structure of human writing—to the point where they are capable of so accurately identifying repeatable patterns in data that the responses they generate cleverly mimic human responses, often doing it so well that they are indistinguishable from human creative outputs.

If we continue down this path, is it possible that writing, once the quintessential expression of human creativity and communication, will eventually be most frequently produced by a non-human intelligence? I am often called upon to provide my views on the implications of AI technology on our lives. The discussion mostly veers towards the usual tropes of artificial general intelligence, the fear of bias, and the concern that, at least in the near and medium terms, the jobs that many of us are doing will no longer be relevant. The more serious worry, in my view is that we stand to lose the very essence of human communication and knowledge.

Yet, I retain hope in the potential of this technological evolution. I believe that AI can augment our abilities and help us navigate the global ocean of information more effectively. But for that to happen, we need to strike a balance between leveraging AI’s strengths and retaining the humanness of our knowledge and communication.

Only if we manage to do that will we be able to shape a future in which technology complements, rather than supplants, the human experience.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

error: Content is protected !!