Life After Language
ribbonfarm.com
ribbonfarm.com
I think, and am basing this off of nothing objective, but it feels like the vast majority of content today is written by people who spend, frankly, too much time online. I think the fact I’m writing a comment about this says that I, too, spend too much time online.
The bubble that the tech world and this website in particular differs so much so from reality that I don’t know what to make of the articles that front page here.
All the twitter doomsaying, trump will never be president, all entirely wrong. I don’t know what to make of it - maybe I should log off for a while
These days I schedule my work around the weather. Few things bring me as much happiness as a day in the sun. I know it has been a good day when I have not touched my laptop once.
It's a bit silly, and still very bare bones, but I just like the phrase "touch grass", and this is my effort to reclaim it from the depths of derisiveness.
And actually you can combine both. Sitting with your laptop on the grass (in the shade), to get the outside feel, but still work done.
As with many things in life, though "go touch grass" isn't actually about touching grass, touching grass really is a good thing. (Well, except that it makes me itch all over the area of contact. Still worth it.)
At least, that's what works for me, sometimes metaphors are to be also taken literal.
I can't afford to go to restaurants all the time, and I don't like bars or the mall.
I'm shy and I don't do well around strangers, and even when I do meet new people 99% of the time we don't have much in common, so it feels like a waste of time.
I'd much rather surf the web... at least there I'm learning stuff, and I can communicate with people who I actually have something in common with.
Some things, like spending time in nature are nice once in a while, but there's no nature near where I live, and even if there was I'm often not in the mood to go.
Going to restaurants, bars or meeting strangers isn't what I meant. It's still the artificial human made world. Spend time outside the city, hiking, boarding, climbing, running or just enjoying nature... That's it. Finding like-minded people will come by itself.
Oh, and beeing active is just awesome. Pushing the body to certain limits is just so important for my mental health. I am a complete different person, if I don't do sports for a certain time.
[1] https://www.reddit.com/r/slatestarcodex/comments/9rvroo/most...
I worked in conventional ELV systems related engineering before getting into tech. I feel like people outside high-tech life are so much behind in many aspects of life - the progress made by tech on the lives of high-tech people are yet to reach the ordinary person.
There are so many non-tech people out there we look down but in fact they are the probably 98% of humanity and their reality define how we as humanity evolve.
Like finding out your 55 year old Aunty has secretly been using LLMs to automate part of her job is real.
“High tech” people are like trekkies, they’re just more into it.
Note: I'm not stating that people online don't watch TV at all -- just that they don't derive the same sort of enjoyment from it that normal people do.
It'd be great if I could afford to travel, but I can't... and, anyway, traveling itself has many downsides, and you can burn out on that after a while.
After the first 5 times you have those conversations, you can predict the entire night's conversation before they begin. I'd love to have offline conversation that was almost anything else.
The online world brings one useful thing to the meat world: organization of meetups for folks interested in a bit more… take Jazz, for example… thoroughly possible to organize a jazz jam in a suburban area thanks to the net.
* despite having been indoctrinated for years that the genius of the States is that "anyone can become president"?
Genuinely curious.
Not on Twitter, or Mastodon, or anything - just real people. Hell, you can even get away with joining a small discord server and joining a voice chat there. Anything but the large, hype- and drama-driven user generated content ad farms.
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?
> Now that a future beyond language is imaginable, it suddenly seems to me that humanity has been stuck in a linguistically constrained phase of its evolution for far too long. I’m not quite sure how it will happen, or if I’ll live to participate in it, but I suspect we’re entering a world beyond language where we’ll begin to realize just how deeply blinding language has been for the human consciousness and psyche.
I think the LLM revolution tell us something else - it is language that crystallises and captures almost all the human intelligence. By training a random initialised transformer on language you get chatGPT. Doing the same to a baby leads to a modern human. It was not the brain or the transformer that are intelligent, it is language running on them that is. Language is a self replicator, but it needed humans, now it works on LLMs too.
In the long term I expect AI will discover things that are hard to express in our language and probably need to develop and teach new concepts to us. And of course AI can help create languages, it can do that quite well, if that's our thing. We can develop private (hyper-)languages and have it translate back and forth.
This used to be a common position held by the likes of Wittgenstein, but the biological reality is probably different [1]. Language is probably like the shadows in the Allegory of the Cave, a representation of some thing not to be confused for the thing itself.
[1]: https://mcgovern.mit.edu/2019/05/02/ask-the-brain-can-we-thi...
My point was the capabilities we attribute to LLMs actually come from the training data, not the model. Any model would do, from 10B parameters to 500B, with cosine embeds, rotary or AliBi, even RNNs can do it today. But the data - that is precious, the source of all these "emergent" abilities. We are wondering at the mysteries of the transformer, but the mystery was hiding the training set.
Let's look at the problem from the point of view of language as an evolutionary system of ideas. Like scientific papers and citations, that kind of stuff, leading to AI and human progress. That's where our progress will come from - the evolution of ideas.
You're talking about "memes", in the original meaning: https://en.wikipedia.org/wiki/Meme
In reality we think not just in words but in pictures, sounds, abstract unsymbolized concepts. We also just do with our bodies and just watch with our hearts.
Image models capture some of that. We don’t have a way of capturing our unsymbolized thinking or doing yet. Expect lots of video data to be used to train AI in human body movements.
Unsymbolized, and worse subconscious, thinking will be harder. My assumption is we will have AIs which can learn directly from the world rather than via curated human culture by that point, so won’t need it.
Same with training it on programming languages, which have a similar underlying form.
What I've found though is that this "reasoning" starts to get exposed as shallow fairly quickly. It can only "backtrack" through logical relationships a couple levels deep. E.g. I give GPT4 a Rust program with a borrow checker problem and ask it to "fix" it and watch it flail around in circles for 3 or 4 tries before just producing complete nonsense because it gets lost.
It looks like human intelligence like you say, but it just falls over after a few short steps.
Yes, most humans will do the same thing. Except unlike an LLM, humans are able to sense when they get lost and attempt to sometimes retrace or correct their steps. But I think there's something intrinsically missing in LLMs that prevents them from doing that -- they don't "know" why they arrived at some step, and don't seem to be able to retrace and fix. Ask them to fix something they get wrong, and they start to degrade in quality quite quickly, I find. Self-awareness/consciousness may be the kernel of that missing piece.
I think maybe these systems might be more useful once they're tied to an actual rigorous logical deduction system, and "taught" how to use them. Let the LLM "read" the prompt, recognize that it needs to do some logical constraint solving or deduction, etc., create the right terms or horne clauses or whatever in that subsystem, query it, and then let the LLM in turn produce the textual response.
This sounds like it's leaning or leading toward Julian Jaynes bicameral mind book or ideas?
It is not incoherent if you believe art to be an expression of the human condition. You still need humans in the writer's room to know what they want the audience to get out of a show. That is not something that ChatGPT can understand.
Also, maybe people should have some feelings "about the economic fates of career writers." Who are we even building this hyper-efficient economy for?
These think pieces that subscribe to an unspoken underlying technological determinism are so disgusting to me, they have a really narrow and pathetic view of what it means to be human (it mostly boils down to "economic agent") Half the time I wonder if the authors themselves even have requisite humanity, and I also wonder if the obsession around digital technology is partially driven by our having made machines of ourselves in the first place (all we care about is economy, work, etc. We've lost the old notion of the "human spirit" at great cost).
More people in power and decision making positions need to start broadening their reading lists with Goethe, William Blake, Novalis, and other awakened poets and start letting some of these imaginative ideas about humanity's potential drive their lives more than purely economically motivated crapped out thinkpieces.
My banjo player recently told me our mandolin player describes things in flats because he's originally a trumpet player. My best friend is classically trained with a masters in vocals and couldn't jam with rock musicians because of the way they call out keys.
And Western music is an established, though often debated, language.
I just don't see the difference between any of this. It's still language. Whether, as in the article it's the language a kid invent with their imaginary friend, how I think about and remember the songs I wake up with...language conveys information.
And I don't understand what the author of this piece is trying to convey about language, except perhaps a connotation of the word language I don't understand.
Sounds like the article is predicated on having completely missed the point of talking to another human being?
Perhaps part of the problem is that we are so bad at teaching communication skills that we find it difficult to connect to other people?
Here are a few example prompts that could lead to interesting responses:
"Write a short story about a detective who solves crimes by talking to plants."
"Imagine a world where time flows backward. Describe a day in the life of someone living in that world."
"Compose a poem about the beauty of the night sky as seen from the perspective of an astronaut on the moon."
"Create a fictional conversation between a philosopher from ancient Greece and a modern-day scientist about the nature of reality."
"Design a magical creature that has never been seen before, and describe its unique abilities and habitat."
Remember that "interesting" is subjective, and what one person finds interesting, another may not.
From that point of view, the two examples that seemed the most promising where 4 (fictional conversation) and 2 (time flows backward), but the results I got were anything but novel: the fictional conversation hit the usual tropes, with nothing particularly related to ancient Greek philosophy or to modern science specifically, and for some reason the backwards time world still had meals ordered in breakfast/lunch/dinner sequence. Nice for essays from an elementary schooler (or an AI?), but nothing one would wish to pay attention to read.
On the other hand, at least the backwards time response reminded me of https://www.youtube.com/watch?v=i6rVHr6OwjI , which did meet my criteria for interesting when I first ran into it, so there's that...
(note that "Entropic Time" has multiple possible tangential successors, based on the various details in its interpretation. that's a sense in which it contrasts to the —to my perception— sterility of the AI responses)
Edit: I guess what I'm saying is that I found "Entropic Time" to be a strong move in a Hessian Glass Bead Game, but I guess at this point instead of quibbling about the weakness of the AI moves, I should be impressed that it plays at all.
cf https://i.pinimg.com/originals/67/bc/35/67bc35746db0758b5ab4...
What does he mean by this? There's no real difference between the language in the First Folio and what's published today. A few spelling tweaks here and there, pretty much.
https://www.litcharts.com/shakescleare/shakespeare-translati...
Or perhaps exactly those. Maybe the article author means that's how Shakespear is read by everyone today?
Yes, there is not reason why AI must “speak” in English or any other human language, but any language it speaks in will be lossy in the same way. Fundamentally, all of communication is lossy including our computer programs. This is why things like undefined behavior exist and also why people misunderstand.
There's a weak and strong version of what I feel is the core idea, "what if human communication becomes largely intermediated by machines".
The weak version is already here, as the article points out. We (at least we, chronically online people) already talk in memes and gifs quite a lot. And how AI will shape it is worth thinking about. I saw that happening on real time on TikTok with people using alternatives for words they feel get their content deprioritized, like suicide, murder, sex, etc. A literal AI model changing language.
The strong version, though, is purely star-trek/Asimov level of speculation. Let alone the same neighborhood, current technology is not even in the same country as that.
The article kinda mixes both things back and forth for a bit, and then fully dives into the scify version, which makes me wondering if the author genuinely believe they will see anything like this. If they do, boy will they be disappointed.
Now, assuming it's all fun scify speculation of a future we'll never see, but hey, might happen, I have some problems with the ideas, but I'll highlight two.
> Plenty of other species manage fine with simpler languages or no language at all.
No they don't. Only species doing well right now are the ones we breed or our parasites. We can potentially erase almost all non-microscopic life forms by accident, let alone if we really put our minds to it. Humans have effects on a geological scale, that's a very select group of species.
> And it is not clear to me that intelligence has much to do with the linguistic fabric of contemporary society.
Kinda see the above comment? I don't think we even have a concept of intelligence that isn't somehow based on language, even if that language is not necessarily some form of spoken language. We certain can't evaluate intelligence without it. And as far as I know, humans don't develop language without socialization.
Sure, you could envision a future where the machines do the socialization, but I think that would be a civilization that would be too fragile to exist for much long.
Another interesting point is the idea of Logos, Word, Speech as the divine attribute par excellence in some religious and mythological contexts.
That said I am not sure what the author wants to optimize. I mean if AI can mediate meanings and intent, that sounds interesting. If it is to AI have their own chats, I doubt that is useful from a human context.
The basic meta mistake in many similar arguments that I since heard from others in the space is exemplified in this:
> That’s… not how technology works.
It's... not about technology, tech is not the end goal, human is. Tech serves human, and if it hurts human guess who should take precedence.
Some questions are easy to dispense with, though: should we hold back the progress of ML because of the job-security concerns of some Hollywood writers? Answer: "No."
Gonna throw up a nice big [Citation Needed] flag on that one.
A rising tide lifts all boats in the end. All of them. No exceptions.
You're putting the cart before the horse. ML is what will enable UBI. In the best of all possible worlds, we will look back on our present economy a hundred years from now and shudder, just as we currently do when the conversation turns to company towns and antebellum plantations.
That won't happen unless something changes, and... well, here we are, watching it change.
If zero-sum gamers in government don't "regulate" ML to death first, that is.
It's an entertaining thought.
In a weaker sense, long emails and single bullet points may have adjunctions between them. (every single bullet point may be expanded to any of a set of long emails, but one suspects that many long emails are incapable of being summarised by single bullet points)
The cartoon is interesting in ways that https://news.ycombinator.com/item?id=35840240 fails to be.
Fluent content-free drivel is now essentially free, whatever that's worth.
this is a play on a Chinese media quip attributed to Xi Jinping that says "changes not seen in the world in 100 years" https://www.strategictranslation.org/glossary/great-changes-...
we have just found (in the 20th century) something that I think is as BIG as writing itself.
Writing taken as the underlying technology behind (or under?) the concept of LAW itself (hence, a foundation for the notion of 'rule of, by, from, LAW" and everything and anything founded on this notions has become candidate for revision and re-evaluation).
all because we invented a way to write that can change itself; this is usually better known as computing. I write some words that say how some other words must change.
It really is all substitutions of arbitrary symbols (or sequences of symbols) for other symbols (or sequences thereof) all the way to the bottom. At the bottom there's a first number and a huge debate (and confusion) around whether this first number is one or zero (LOL).
Couple this with some form of advanced VR, and human-to-human collaboration will make everyone look and sound either: a) the way they would LIKE to be presented to the world, or b) an overriden version of how the OBSERVER would prefer them to look and sound.
Either way, it's terrifylingly isolationist.
I found myself persuaded into the opposite point of view with the passage of each intellectually bankrupt paragraph.
Congratulations, I’m thoroughly dissuaded of any such conceits by the end of reading this, I want my 5 minutes back.
Tell me a chatbot wrote this without telling me a chatbot wrote this
"There's a new kind of dystopia that's seeming increasingly plausible. A post-literate society where the written word has been debased by LLMs and where those who can focus and consume long form prose will increasingly be unable to effectively communicate with those who can't."
Even the fact that the author held the conversations they had with a LLM higher than human conversation proves exactly this point.
One step further, how are humans going to exist when AI handles everything? Why communicate at all if days can be passed in a drug induced stupor of bliss?