 |
Amazon link |
Tyler Cowen, from the blog ‘Marginal Revolution’, suggests that we should be writing for the AIs, the large language models (LLMs) that will one day serve as our digital custodians.
The reasoning is straightforward: our descendants will have neither the time nor the inclination to sift through the mountains of words we leave behind. But the machines will. They will be the ones reading, processing, and perhaps even presenting us to the future. It’s an intriguing notion, but one that raises an immediate question: why bother?
Human relationships are defined by bonds of personal necessity and affection - family, friends, lovers, comrades-in-arms. Beyond this circle, we lapse into mere social obligation, and even that wears thin.
If I had the chance to meet a sixteenth-century ancestor, what exactly would we talk about? The price of grain? The relative merits of different plough designs? His pious peasant worldview would, at best, be a short-term curiosity; at worst, just tedious.
Why, then, should I expect my distant descendants to care for my accumulated wisdom on obsolete telecom network architecture, frankly naive political philosophy, or the minor ups and downs of twenty-first-century Britain?
But perhaps that’s the wrong framing. The real issue is not whether anyone will care, but whether anyone will even feel inclined to. The function of human relationships is mostly to exchange knowledge, provide advice, offer emotional support, and help each other navigate the labyrinth of life. Yet all of this will be, increasingly, the domain of more competent and amenable artificial systems.
Give AI another twenty years to develop fine-grain personality modeling and soft-touch humanoid robotics - and the role of wise old grandparent, experienced mentor, or sympathetic friend will be performed more effectively by an exquisitely tailored machine than by some stilted, digitised reconstruction of a long-dead ancestor.
The living barely listen to their forebears now; why should we expect the future to be any different?
Perhaps we should adjust our Wills. State that our digitised personas may be edited, altered and upgraded, tailored to become a preferred companionate personality for our remote offspring.
In this way allow our individual, historic selves to be subsumed into the collective subconscious of improved and optimised humanity: a Jungian prospect reminiscent of Peter Hamilton’s Edenists in the Night’s Dawn trilogy. The self, once rigid and finite, dissolves into something fluid, continuously enhanced - a digital ancestor no longer preserved but reimagined and repurposed.
Cowen’s vision, then, is not without irony. Yes, the AIs will read our words, digest our ideas, and perhaps synthesise them into some grand collective narrative of human thought. But what remains of us in that process? Do we become mere echoes, endlessly refined until we are no longer distinct from the computational matrix itself? Or do we vanish into irrelevance, our carefully curated personas serving as little more than scaffolding for something newer, sharper, and ultimately indifferent to its origins?
If we are to write for the future, we should do so not out of a desperate hope to be remembered, but because we have something genuinely worth saying. Process not outcome: whether it is read by man or machine is, in the end, incidental.
