Remix.run Logo
▲ reworkal a day ago

> No, Claude flew, and danced, and sparkled near that finish line.

Side note, anthropomorphizing your stochastic parrot sort of distracts me from your argument. Like sure guy might be a genius but as someone who knows nothing about you, reading this is very off-putting.

This is like my own opinion.. not saying what he wrote is bad, hope its real TBH

▲munchler a day ago | parent | next [-]

It might be off-putting, but perhaps it’s time for you to face the fact that Claude (and other high-end models) are way past being stochastic parrots.

▲chilipepperhott a day ago | parent | next [-]

I think you're missing his argument. Claude is a tool, not a person. It's just as counterproductive to anthropomorphize my hammer as it is to anthropomorphize an LLM.

▲munchler a day ago | parent [-]

Claude is a tool, but recent evidence suggests that it is also intelligent. Your hammer does not exhibit intelligence.

Note: I'm not anthropomorphizing anything. I think the best description of Claude right now is that it is an alien, non-human mind.

▲aerhardt a day ago | parent [-]

I have no problem in accepting that LLMs are vastly more intelligent than humans in many areas already, but to me the concept of mind involves "beingness" and maybe even sentience, and I don't believe at all that these machines have any of that.

"Alien" is an acceptable term in the sense that these things reason in some very different ways than we do, but due to cultural norms, to me it also brings up notions of beingness. The xenomorphs in the alien movies, vile (and fictitious) as they may be, have a certain ontological quality that the LLMs do not.

▲munchler a day ago | parent [-]

I think your position is reasonable, but even if you're right, the "mere stochastic parrot" position doesn't hold water anymore.

▲aerhardt a day ago | parent [-]

I've personally never held that position. I always thought that even in the minimalist case for AI, "stochastic parrots" would have mind-bending implications for our understanding of verbal reasoning. I think that we're well past refuting the minimal case from a couple of years ago, anyway.

▲mbgerring a day ago | parent | prev | next [-]

Be specific, in what sense has the actual method of action of LLMs changed, according to your understanding?

▲munchler a day ago | parent [-]

I don’t think it has changed. The difference is in the model size, complexity, and weights.

▲akomtu 20 hours ago | parent | prev [-]

AI is a swarm of stochastic parrots, meaning they use pattern matching as their guide. With enough number of such parrots we apparently get an alien form of intelligence. In other words, don't underestimate stochastic parrots.

▲timmmmmmay a day ago | parent | prev | next [-]

how many unsolved math problems does it have to solve before you drop the parrot thing

▲luka598 a day ago | parent [-]

well, if you make the parrot big enough and teach it enough skills it perhaps becomes usefull, but it still is just a parrot

▲cman1444 a day ago | parent | next [-]

I guess we're all parrots then.

▲FartyMcFarter a day ago | parent | prev | next [-]

Yes, it looks like parrots have evolved and become extremely intelligent parrots.

▲chpatrick a day ago | parent | prev | next [-]

What proof of the Riemann hypothesis is it parroting exactly?

▲azan_ a day ago | parent | prev [-]

Yes, if you stretch your definition far enough everything will be a parrot, including humans.

▲903543958568 a day ago | parent | prev | next [-]

You luddites sound more stupid with each published proof.

▲kyle-rb a day ago | parent | prev | next [-]

He said "Claude flew", which fits perfectly with your stochastic parrot framing. He's not anthropomorphizing, at worst he's ornithomorphizing(?)

▲jimmysnuka a day ago | parent | prev | next [-]

Can you provide a convincing argument that your brain is not also a "stochastic parrot"?

▲zeroonetwothree a day ago | parent | next [-]

If it is I don’t see how that is an argument against the original claim :)

▲FartyMcFarter a day ago | parent [-]

It's only an argument against the implicit derision.

▲gloosx a day ago | parent | prev [-]

hmm, my brain continuously experiences and interacts with the real world, rather than briefly coming into existence to process a virtual request before vanishing into eternity?

▲preg_match a day ago | parent [-]

It kind of sounds like the only limitation or distinction then is processing speed. The tick-rate of intelligence. LLMs have a very slow tick rate.

But, if I had a camera or something and constantly fed that to a model inside a harness, is it then a brain? I mean that’s basically exactly how the brain works - it gets electrical stimuli from a bunch of places, makes some decisions, and then spits stuff out. The loop is just much, much tighter.

▲gloosx 14 hours ago | parent [-]

When an AI gets a body and starts continuously experiencing and interacting with the real world, initiating its own actions without waiting for someone to prompt it, I'll accept that we are fundamentally the same.

The distinction is whether it can generate its own reasons to act based on real world stimuli, rather than having every interaction initiated by someone else. And it's not just about hooking a camera to a harness to provide vision, it is also hearing, smell, taste, touch, thermoception, nociception, proprioception, vestibular sense, interoception, hunger and thirst, social stimuli, etc. In order to deal with all these stimuli feeds your harness needs continuous memory and learning ability and internal drives. It needs to be a continuously operating, self-directed intelligence, which it isnt right now.

And we also need to remember that rather than explicitly training a brain to recognize objects, understand language, and solve problems, brain has the mechanisms to develop those capabilities through interaction with the world starting from roughly 800 MB of genetical information to produce 86 billion neurons and hundreds of trillions of synaptic connections. That development is not a direct assembly from some detailed blueprint, it is generative, it does not separate training from inference. I think all of that combined explains why we call LLMs a "stochastic parrot" and brain isnt merely a system that produces plausible responses to inputs. Its a continuously developing, embodied, autonomous process that actively interacts with the world, generates its own motivations, and learns from the consequences of its actions.

▲azan_ a day ago | parent | prev [-]

> stochastic parrot

This has to be satire, right?

▲dude250711 a day ago | parent [-]

Could be a time traveller from 2024.