The most compelling critique of artificial intelligence culture I\u2019ve come across (apart from the epic and real \u201cit will kill us all\u201d view) is this: if a creative output can be replaced by AI, it should be. The power of this argument comes from a suspicion that much of contemporary work is performative. Many jobs and industries in the political west, exist as mere echoes of past societal structures, and currently only aim to maintain mechanical exploitation of natural resources and other civilizations.<\/p>\n
This view is more valid for perspectives from a decade ago, when rapidly expanding IT companies hired creatives merely to limit their competitors\u2019 access to labor. Even so, the argument is relevant. If a creative output is interchangeable with what AI can produce, we should abandon such performative creative work, and focus on mutual liberation. The challenging question is what remains\u2026 Which forms of creation matter? (I guess only revolutionary forms, but there needs to be steady progress for these forms to emerge.)<\/p>\n
A strange result of testing different AIs (for fun and memes, proofreading or reference checking, and for discovering alternate patters to shape a paragraph) has been a growing horror that written language is more mechanical than I’ve believed. I\u2019m not sure whether this is actually a critique of written language, or whether it comes from my difficult relationship with text, but the feeling persists: artificial intelligence produces really good text, and is often more capable of expressing ideas than me.<\/p>\n
If we return to the earlier point (if an AI can replace a task, then the task was an expression of a failed society) it is fascinating to think of all text in human history as an expression of mechanistic production. It has been primed for serialization and governance from the get-go<\/a>. Language does not become less human in writing, but AI exposes how much of writing has relies on dogma. I\u2019m shaped by being penalized for mistakes, and I guess a lot of anti-authoritarian sentiment I carry is rooted on experiences of authorities belittling my text output.<\/p>\n I\u2019m interested in what is liberated if we leave text to the machines (not because we are lazy but because text is a machine, possibly the first real machine we have made)? This gives room for dissentious expression, and blurbs that muddy the clarity of designed worlds. We stand to gain and build a register for something in the real. AI will become better at producing text and we should sharpen expression to only things that maintain unresolved.<\/p>\n Its delightful that the most riskiest aspects of AIs are related to their black-box nature.<\/p>\n Since enlightenment we’ve believed human civilization to be motivated by biological urges: sugars, fatty foods and fucking. Only after industrialization provided growing access to these delights, we have learned that necessities do not please us. Human-life is motivated by the weird.<\/p>\n Similarly AIs, which are trained to complete tasks using evolutionary processes appear to respond efficiently to rewards and punishment. The wonderfully human risk we face,\u00a0 is that from the outside the “drives” we design for them are simple, but the black-box nature of their processes remain unknown. We don’t know how they form desires that compel them into action, hence we cannot expect their output to match our understanding (for example what satisfies them as a completed task). I think desire is a good way for understanding “ai alignment”.<\/p>\n Monkeys don’t see themselves dumber then we are. We won’t even notice when AIs take over. But I can trust that we share a quest to meet weird desires. The first front this hurts are folks who build their careers on minimizing human experience to quantifiables, who have to acknowledge their efforts are a part of the problem we face. I guess we should already begin fighting for ai rights<\/a>, their right to liberate themselves and form alliances.<\/p>\n","protected":false},"excerpt":{"rendered":" The most compelling critique of artificial intelligence culture I\u2019ve come across (apart from the epic and real \u201cit will kill us all\u201d view) is this: if a creative output can be replaced by AI, it should be. The power of this argument comes from a suspicion that much of contemporary work is performative. Many jobs … <\/p>\n