Sounded controversial to me when I first encountered it, but apparently accepted by modern philosophers as a legit argument. (That language is necessarily a public/communal thing.)
Ok, that is interesting and I can kind of see it, but this is basically assuming you already have acquired language and talks about the actual act of using language. Otherwise it's a bit of a circular argument: consciously deciding what a word means, as opposed to intuiting its meaning from experience, context and memory, requires being able to consciously structure and organize thoughts. And for that you need to be able to categorize the thoughts, which requires labeling them, which already presupposes a rudimentary form of language.
(Also, if I were to "make up a word with a meaning", use it in a sentence I write down, then read it again the next day, then there is something stopping me from changing my mind on what it means: making sense of my own past thoughts. But this is a bit of a cheat because "past me" and "current me" are arguably two different entities)
The problem I have with Wittgenstein (and many philosophers) is that they rarely, if ever, seem to leave the domain of language and semiotics to consider the world outside of the signposts that the signposts actually point to. They're having too much fun with their language games.
Meanwhile, Sacks is always trying to stay grounded in what we know from research and case studies. For example Seeing Voices describes accounts of the prelingual deaf who acquired sign language later in life, and therefore were able to describe what their experience of acquiring language was like (since childhood amnesia[0] did erase the memories of this process). I can't fully recall the details on the spot here, but living without language essentially is like suffering from an absence of mental "scaffolding", making it impossible to compose ideas together to build cohesive, complex lines of thought.
So sacks also concludes that we need other humans to acquire language (really, to become human in general), but it's more of a matter of bootstrapping: in order to acquire language you need a dialogue, and you can't have that if none of the participants of the dialogue posses a language!
> So sacks also concludes that we need other humans to acquire language (really, to become human in general)
I do agree with Sacks' conclusion, having first come across a version of it in Hegel's dialectic--the dialectic of self vs. other. Or in my interpretation, of personhood and relationship. i.e. Personhood is forged through relationships with other persons.
Do you think this applies to AGI? That it needs to bootstrap itself by communicating and relating with other intelligences? This, and related concepts like free will, tend to be ignored in the scientific literature on the subject[1].
[1] https://arxiv.org/abs/2308.08708 Consciousness in Artificial Intelligence: Insights from the Science of Consciousness
That sounds plausible. On the other hand, it might be another case our tendency to judge intelligence by the already existing intelligence that we know of, which is us. But we probably shouldn't assume that that is the only form of "intelligence" that exists.
Related to that: these types of discussion tend to get rather fuzzy because most of the words involved ("intelligence" especially) aren't nearly as neatly defined as our gut feeling tends to make us believe. And depending on which interpretation people pick they can wildly talk past each other, or even confuse themselves by not realizing they're switching between which aspect of intelligence they're talking about in the middle of their argument without realizing it.
Which is precisely why I don't really feel the urge to go deeper into what might be needed for AGI right now though, apologies. I think we maybe should put a bit more effort into understanding humans first.