The Losing Craft

We have breached an event horizon and there is no exit. First they came for the writers, the designers, the developers. After the outsourced lower tier labor - the reviewers / moderators, assorted contractors had been dispatched, the most vulnerable to the frenetic hype machine of efficiency were those whose professions were already tangentially aligned with the easily reproducible - the templated, the scripted, the predictable, the binary. 

The machines would not be a replacement, they said, but a counterpart, a companion, a ‘co-pilot’. They would run off of huge datasets - billions upon billions of entries to the canon of the human mind and experience, creativity scraped and stolen, converted into matrices, trained, organized and algorithmically stacked into neat probabilities. 

They would have UI that resembled a human interlocutor, taking prompts of human words and rendering output that can produce an answer to any conceivable inquiry or problem in seconds. Of course, most human users would never know that the machine does not actually use words, but tokens and calibrated weights on inputs, numerically converted within a vast transformer. And the real problems were not really solved, but re-arranged, re-packaged and delivered back to us as fragile veneers of solutions. We left more confused, more indebted, and less capable. 

The most pernicious threat is one that resembles progress so convincingly as to estrange us from one another, from our own minds while presuming to unite us behind some common goal. Collectively, the agreement to sacrifice our uniqueness, our capacity for originality, our specializations, our purpose to a corporatist fantasy of a de-humanized labor economy has the potential to be among our most catastrophic social failings.

The billionaire corporate class force feeding AI to the public with the specter of an AGI as a utopia of efficiency and opportunity is a real time economic, environmental, and human rights catastrophe. 

Compounding the macro-economic uncertainty where thousands are losing jobs in political purges, costs of living rise and rise, and surveillance capitalism underwrites every technology and tool, you now are expected to passively outsource your thinking and privacy to LLMs as a matter of company policy and social protocol. Ignorance through complete techno-dependency is our future state. Yet here we are, and we must engage with our occupiers. 

Large Language Forces

The impossible standard of being both compelled to use LLMs and simultaneously discouraged or shamed for using them is also one of the vexing conundrums of our era. You can’t have an ‘AI assistant’ in every application and also characterize the usage of AI assistants as manipulation or cheating.

Is using a commercialized, mathematically powered plagiarism tool a form of academic or social dishonesty? Maybe. Maybe there’s a deeper duality in a mechanism that aspires to harness all of human knowledge at any cost to be summoned on demand and the near total surrender of original thought and intellectual ownership to this mechanism.  

The saddest part of this transformation is that we recognize the dystopian absurdity of it all. LLM generated content, especially writing, is (mostly) unmistakeable, hideously redundant, verbose, and generally very uninteresting. But it is getting better. Images and video can also be spotted, but it is getting more difficult. Writing, art, design have always been turbulent domains but are now crafts in true peril. Software development, once a model of professional resilience, is an endangered species - a craft whose value is degraded from within by technocrats, hemorrhaging capital and burning the planet to coerce you into a belief that losing your livelihood to an LLM is actually in your best interest. 

The pursuit of making something yourself has little value in an era of prompt generated resources. Future generations may also lose the desire to create for the sake of creating - for the pursuit of exploration. When a perfect solution is available on demand, why actually learn to do anything? The scaled contracting of entire professions to LLMs will do the most damage intellectually and economically. The mass de-incentivizing of learning and total emphasis on quick solutions as millions lose jobs to automated systems prematurely deployed to replace human labor in the name of ‘efficiency’, coded in the language of ‘liberation’ and ‘innovation’, is the ultimate conceit of the new gilded machine age. 

While the focus lies on the technology itself, LLM tools replacing us isn’t the core problem, but rather how we collectively choose to stop doing things - both through strong external incentives and personal opposition to friction. Rarely is an external force singularly responsible for altering the direction of human behavior; instead we elect to change our mindset and adapt to environmental shifts. The tool is just an object; we assign it meaning, deify it, and shape our institutions and modalities around it. Reacting to the external force isn’t always negative, but we can become so consumed and credulous by the ‘object’ that we allow ourselves to be subtly directed and eventually irreparably controlled without much resistance. 

The ability to choose and to do are powerful features of the human mind. With awareness, we can counter the most pernicious effects of the AI era. We can choose to continue to think independently, reflect, create, and preserve what makes us unique and capable. We can choose to what extent we allow the incredible progress in computing to act on our behalf and make human life more equitable, or irrevocably advance the self-engineered displacement.