3 ms·
I see all this as the evolution of language, because that's how ideas are preserved and replicated. Ideas evolve over time. They form an evolutionary system, mo
by visarga 3y ago
I see all this as the evolution of language, because that's how ideas are preserved and replicated. Ideas evolve over time. They form an evolutionary system, moving much faster than biology. Humans and AI both rely on the corpus of language to learn and apply ideas.
We're no longer the only way ideas can replicate - LLMs have joined the party. Language gained a new mechanism for self-replication. We should focus less on humans vs AIs, and more on the language corpus itself, the true source of intelligence.
This corpus was built through experience, slowly and collaboratively, by billions of people in parallel. No single human could have done it alone - it's smarter than any one of us, the product of our collective work and luck.
When you train an LLM on it, the model becomes generally competent. But did the competence come from the model or the data? Should we obsess over architectures, or look to our datasets?
- diego_moita 3y agoI've heard that before, mostly on Ray Kurzweil talking about singularity. But I have a problem with this usage of the "evolution" and "evolutionary system" expressions. I believe that it entails "improvement", that it is judgmental. I think it is self-centred and presumptuous to imply that the world improved toward us. I don't think Darwin used the expression "evolution" with a judgmental meaning. He only meant that it was an adaptation to existing environmental conditions.
- visarga 3y agoIt's just self replication. Ideas that are useful in some way get replicated, other ideas forgotten. They travel a lot and mutate. They all depend on human interests, but exist beyond individual humans.
- satvikpendem 3y agoNice to see that the idea of Richard Dawkins' meme lives on, fittingly enough.
- TheOtherHobbes 3y agoAnd failure to adapt leads to extinction. AGI has the potential to be an improvement on humans because it has the potential to be relentlessly rational, reality-aware, and self-correcting. We don't - most likely can't - operate like that. Although we like to flatter ourselves that we're a rational and intelligent species all of us live in an unaware haze of myths, factoids, media-seeded fake narratives, self-serving fantasies, and outright reality denial. We have a rational process called "science", but politically it lives in a special box. If it starts to interfere with established power hierarchies it's shouted down. And hardly any humans science at all. Current AI is just a warm-up for what happens next. Metaphorically it's the equivalent of 8-bit computing. 32-bit AI will be entirely different, very possibly incomprehensible, and potentially far more threatening.
- ChatGTP 3y agoIf AGI is something that becomes real, I don't think it would be recognizable to us for long, I don't think it will be a type of steady state even like humans, reading books to it's chidren in the cloud and get really good at coding and similar tasks, and improve itself infinitely (into what?). I think life has a special property in that it is constrained, so it exists. Imagine being able to become pretty much anything, what would you choose? We can't even comprehend what a system like you describe would evolve into, but I don't think we can even fathom it, it's not going to be like some ultimate task master who gets 18 whole in ones at gold on Friday and take all our jobs, for some reason ? What I think we're doing is applying our problems and the things we'd want (infinite self-improvement) because they're things we feel would be an advantage to us. A robotic system though? Not sure it would care. It seems like AI is the ultimate end game, but I'd imagine if we survie the AI era, we'll be past it pretty farking fast.