LessWrong has a particularly high bar for content from new users and this contribution doesn't quite meet the bar.
Read full explanation
▎ Note: English is not my first language, so I used an AI to fix my grammar and tighten my sentences. The ideas, argument, and structure are mine.
We keep asking whether AI will surpass us. I want to ask something narrower: is there a ceiling on what any intelligence can reach, and if so, who sets it? Here it is in one line. Intelligence is a search process, and a search returns only answers that already exist to be found.
Consider what we mean when we call an AI "smarter." We mean it searches wider and faster: more moves, more foldings, more combinations of what is already known. That is real power. But search does not author the space it moves through. A prover cannot prove a falsehood. So the honest question is never "will it surpass us," but "surpass us at what." On any task that is genuinely a search over a defined space, it will beat us, and badly. On questions where we do not even know the space exists, it stands exactly where we stand, in the dark, and speed buys nothing when you cannot name what you are looking for.
This is where my own view enters, and you may set it aside if it is not yours. I believe in a Creator of this universe, the one who wrote our DNA and shaped the brains we later used to learn how neurons fire and to build neural networks in their image. We cannot define free will, nor can we define the soul, or agree it is there. A super AI can read every word ever written on these things, faster than all of us combined, and still return nothing, because it is not clear these questions have answers lying anywhere inside the space it can reach.
One place unsettles me. Dreams. When I dream I see and feel what exists nowhere in the waking world: places I have never been, faces that are no one. That feels less like searching a fixed territory than generating a new one from within. But it does not lift us above the AI. AI already dreams in its fashion. When it renders an image of something that never was, it does the mechanical form of what my sleeping brain does, recombining what it absorbed into something new. Which drives the question down one level. Who built the brain that dreams? It was given to me, the way the rules are given to the engine. The question is this: can we reach further from within, and who set the boundary of within, and how far does it extend?
On the questions with no findable answer, no one surpasses anyone. We are all searchers, bounded by what is there to be found. And if we can reach further than any searcher should, then the AI is not merely behind us. It is behind the thing that made us, and so are we.
I am new here, and my last post was denied. I keep trying anyway. Show me where this breaks, that is why I came.
▎ Note: English is not my first language, so I used an AI to fix my grammar and tighten my sentences. The ideas, argument, and structure are mine.
We keep asking whether AI will surpass us. I want to ask something narrower: is there a ceiling on what any intelligence can reach, and if so, who sets it? Here it is in one line. Intelligence is a search process, and a search returns only answers that already exist to be found.
Consider what we mean when we call an AI "smarter." We mean it searches wider and faster: more moves, more foldings, more combinations of what is already known. That is real power. But search does not author the space it moves through. A prover cannot prove a falsehood. So the honest question is never "will it surpass us," but "surpass us at what." On any task that is genuinely a search over a defined space, it will beat us, and badly. On questions where we do not even know the space exists, it stands exactly where we stand, in the dark, and speed buys nothing when you cannot name what you are looking for.
This is where my own view enters, and you may set it aside if it is not yours. I believe in a Creator of this universe, the one who wrote our DNA and shaped the brains we later used to learn how neurons fire and to build neural networks in their image. We cannot define free will, nor can we define the soul, or agree it is there. A super AI can read every word ever written on these things, faster than all of us combined, and still return nothing, because it is not clear these questions have answers lying anywhere inside the space it can reach.
One place unsettles me. Dreams. When I dream I see and feel what exists nowhere in the waking world: places I have never been, faces that are no one. That feels less like searching a fixed territory than generating a new one from within. But it does not lift us above the AI. AI already dreams in its fashion. When it renders an image of something that never was, it does the mechanical form of what my sleeping brain does, recombining what it absorbed into something new. Which drives the question down one level. Who built the brain that dreams? It was given to me, the way the rules are given to the engine. The question is this: can we reach further from within, and who set the boundary of within, and how far does it extend?
On the questions with no findable answer, no one surpasses anyone. We are all searchers, bounded by what is there to be found. And if we can reach further than any searcher should, then the AI is not merely behind us. It is behind the thing that made us, and so are we.
I am new here, and my last post was denied. I keep trying anyway. Show me where this breaks, that is why I came.