Friendly fyi in case you got the impression he was a recluse isolated in a basement ...
The author "talks to real people" all the time. Examples: https://www.youtube.com/results?search_query=Venkatesh+Rao
I just cannot imagine thinking like this if youre not completely detached from real, normal human interactions.
Your statement makes it seems like you've misunderstood his thesis.
He's envisioning a future where humans-to-humans conversation is more intimate, more meaningful because it's not limited to textual language. Example excerpt from his article: >"We’ll all be like children inventing secret languages for talking to imaginary friends, except they will be real friends."
His attachment to real human interactions is letting him see a future where real interactions can be "post-language". The post-language "secret language" could be images, or dynamic translations customized for the particular person's brain.
How could it possibly be more meaningful. Languages mature over time, bits of other langauges are integrated, vocabularies expand, new forms of expression in the langauge appears, by serious practioners of writing or speaking in the language.
> We’ll all be like children
Oh great. Goo goo ga ga level civilization then. Curious as to how "children" discover the mutual new language and start communicating in it.
> His attachment to real human interactions is letting him see a future where real interactions can be "post-language". The post-language "secret language" could be images, or dynamic translations customized for the particular person's brain.
Sorry, but this is actually a silly notion of the OP. It shows an utter lack of understanding of the role of language in human society. Language is a repository of cultural memory broken down to composable bits.
You’re not being open minded here. Our traditional languages are so inefficient at actually conveying information. Whenever we try to explain very complicated concepts, we always dumb it down to very basic illustrations like philosopher’s diners or ice cream stands explain complex actor model. This is necessary for intuition because using parsing “adult” language at this level is extremely hard.
I think what OP meant is that we will have more of this. LLMs have opened up more possibilities in how we convey information to one another. Why _should_ we stick to English? Why can’t we all converse in generated images, real-time generated videos and illustrations?
For example, instead of having PDFs, slides, Jupyter notebooks or whatever for class material. We will instead of 1000 long line of unreadable and undecipherable prompt to an LLM that will generate that course content that you can talk and interact with?
None of that is child talk.
OP started off on a promising note and one hoped that he would start playing the song of 'vector languages' for machine intelligence -- but alas, he then started singing out of tune.
> You’re not being open minded here.
Unfair. Rather may I suggest you did not read carefully. I did say "civilization". In a civilization, you will need both teachers and students; thinkers, and readers. What language will the thinkers use? "Explaining" is a concern regarding established results. What about deriving new results? (See above)
So we already have examples of discourse in which we alternate natural language with some other, more formal, language: english with formulae (math), english with tabs [less formal] or staff notation [more formal] (music criticism), english with a set of images with formal denotations (memes), etc. Even Montaigne's essays (which were attempts at deriving results, not just explaining established ones) somewhat follow the pattern in that at times they alternate french with classical quotations.
I hope —I think along with eternalban and Hermann Hesse— that we may see the introduction of new combinators, new forms of composition, that are general enough to serve both as explanations of established thought and tools of generating new thought. (cf Aristotle Organon)
> generating new thought
I was beguiled yesterday (reflecting on our chattering mechanical friends) with the idea (new for me) that production of 'thought' doesn't necessarily require a 'thinker'. It's just that we thinkers -- that is the HomoSapiens -- have only known of production of thought as thinkers, and assumed there is 'thinking' involved. (This is what is causing the confusion in some granting GPT the coveted 'thinker' and 'thinking' status.)
If that is the case, then we have misnamed 'thought': It was always nothing more than a proposition, but we assumed it was intimately connected with the experience of thinking.
Genau. I especially appreciate how millefiori glass beads are a physical incarnation of the literary device of quoting other works...
> If that is the case, then we have misnamed 'thought': It was always nothing more than a proposition, but we assumed it was intimately connected with the experience of thinking.
That makes a great deal of sense to me; in that world 'thinking' is like imperative small-step operational semantics, while 'thought' would correspond with functional big-step denotational semantics. The former suffices to produce the latter, but is by no means necessary.
If you thought that this blog post had provocative ideas, you might be interested in reading some of Ludwig Wittgenstein's philosophy of language or Roy Harris's work on integrationist linguistics. These ideas have been debated for a long time.
Yes, I've mentioned Wittgenstein before: https://news.ycombinator.com/item?id=27650343
The older writings about language/linguistics are mostly about observing misunderstandings -- or -- complaining that progress isn't being made in philosophy because people (unknowingly) use different definitions of the same word. They weren't talking about any new yet-to-be-invented machines to bridge the gap of misunderstandings -- and understandably so because they lived in a time before computer algorithms like Large Language Models existed.
In contrast, because Venkatesh Rao is living right now in this LLM world (e.g. him playing with ChatGPT/GPT4/Berduck/etc), he's revising his prior opinion about evolution of communication. This thread's blog post is talking about new possibilities of technology enabling a more direct understanding. Maybe it's a poorly written blog post and maybe his prediction won't happen. In any case, I'm just trying to explain what the article's is about since some readers seem to misunderstand it. (E.g. some readers may think he's a misanthrope detached from real human interaction but his article is actually envisioning future scenarios of the opposite situation!)
EDIT reply to: >Like I said there is a whole debate about the concept of private language that you are missing.
And as I was trying to convey back to you, the previous "debates about private language" are not relevant to _this_ thread's article about future technology scenarios. His article is about predicting a technology trajectory, and not about philosophical debates.
To use this very thread we're in as an example of the author's point:
This thread's article was written by a high-IQ intelligent author. And you, the reader, is presumed to also be highly intelligent. And yet -- you seemingly misunderstood the core idea of what the author was trying to communicate. (E.g. You feel his article is low quality because it re-hashes "private language debates" already covered by previous philosophy). And to use a mirror assessment of the author's writing, he didn't string together the perfect sequence of chosen words to make his point easier to understand.
So... can new technology (e.b. future improved LLMs) ... bridge that gap of misunderstanding ... between what the author wrote ... and your misinterpretation of it?
Neither you nor the author are illiterate... and yet there's a gap between you and the author. Can a future technology help bridge that gap?
Uh... No. Like I said there is a whole debate about the concept of private language that you are missing.
Does he also have a Mensa membership?