Inspired by the divide that emerged in the twentieth century between mass production and handicraft, or between quantum and classical mechanics, I see, in the future, the possibility of two distinct ways of using language developing – ways that differ from one another on a level deeper than that of individual natural-language systems. The limitations carried by language have been a subject of philosophical reflection since antiquity. On top of this, in recent years, in connection with globalisation, democratisation, the growing role of the Internet, and the development of artificial intelligence, enormous dangers arising from the use of language in its unchanged form have come to light. Despite revolutionary changes in almost every field of life and science, the basic mechanisms by which language operates have remained unchanged for centuries. These phenomena, also in the context of the Turing test, suggest that a linguistic revolution may be necessary, if not inevitable. This essay draws much of its inspiration from two quotations from Ludwig Wittgenstein – one about the limits of language being the limits of the world, the other about the nonsensicality of his own propositions.
The limits of my language are the limits of my world.
That every additional language we master allows us to think in broader and more precise ways is, in my view, beyond doubt. The first reason is our familiarity with new words and concepts – constructions that let us formulate thoughts more precisely. Once we learn the nouns hon'ne and tatemae, and the Japanese honorific system, we will look at Japanese people differently – and perhaps at everyone around us as well. The second reason is access to a much larger body of information: being able to read articles about a given event in Le Monde, The Spectator, and Mainichi gives us a broader view of the matter than someone limited to Polish media alone, who may never learn of certain issues at all. Third, as our linguistic competence grows, so does our ability to understand text – lower-level language exams test knowledge of vocabulary and understanding of sentence structure. Exams that test skills at a higher level often also require demonstrating an understanding of what is not said outright – context and authorial intent. Such competencies let us draw more out of every text. And especially in today's world, where we are assailed by text from every direction, the ability to understand language very deeply is a treasure that cannot be overstated. So as we master more languages, we gain more information, understand it better, and process it more efficiently – all in an age called the "Information Age". Of course, thanks to languages the world available to us becomes almost literally bigger – we can fly further, communicate, experience more. Yet in the quotation above, a different problem seems more significant.
There is indeed the inexpressible. This shows itself; it is the mystical.
An ideal language – one capable of expressing with precision everything we think – is a very old dream. According to Jakobson, natural languages differ from one another not in what they can express, but in what they must express. We should not, then, look for an ideal system in any one national language. According to Wittgenstein, language cannot express what lies beyond the world – and it is there that the sense of the world also lies. Accordingly, he calls the whole of philosophy a critique of language. A similar view is expressed in Robert Pepperell's "The Posthuman Manifesto," which argues that most philosophical debates are really linguistic problems stemming from three mistaken assumptions: that language is consistent, that because a given word exists there must exist a "thing" it represents, and that the things thus represented ought themselves to be internally consistent. Plato already spoke of the veil concealing the true, ideal world from us in his allegory of the cave. Perhaps the natural-language systems we currently know, and the necessary stage of transforming thought into sentences, are the "wicked men" who cast misleading shadows on the wall for us. Socrates reportedly avoided writing down his thoughts, claiming that doing so might distort their meaning.
Given the difficulty of describing what language cannot capture, I will focus on the places where linguistic communication fails most. What matters here is that meaning is not carried by language (this is the role of context), but negotiated by the listener within the context of their culture – and that, in reality, the "real world" is, to a significant extent, unconsciously constructed by people on the basis of the linguistic habits of their particular group. Such a picture of the "real world" will therefore be a lie to others. It seems that for as long as meaning remains negotiable within a cultural circle, and people continue to insist on objective truths, the side effect of language will continue to be lies and misunderstanding. The imperfection of linguistic communication in the form we know it is captured perfectly in an observation made by one of the characters in The Plague:
He realized then that he and his interlocutor were not talking about the same thing. For he spoke from the depths of long days of reflection and suffering, and the image he wished to convey had been slowly forged in the fire of waiting and longing. The other, meanwhile, pictured only a conventional feeling, an off-the-shelf grief, a mass-produced melancholy. Kind or hostile, the response was always wrong, and in the end one had to give up trying to talk about it at all.
Words such as love, fear, and happiness are so blurred in meaning – used even by a single person to describe such wildly different states – that it is hard to believe they are among the most important words for our survival as a species. Of course, attempts to make feelings more precise have been made for centuries: platonic love, euphoria, the English dread. Yet it is hard to shake the fear that, in time, these words too will be blurred and flattened by poetry, hyperbole, lies, and the urge to manipulate. Wouldn't it be simpler to be able to just "show" our interlocutor how we feel? This problem is especially visible in today's world – polarised, seeming to stand perpetually on the edge of an abyss, a world in which any spark of conflict could end in nuclear annihilation.
Online, it is not hard to find arguments between believers in the left and believers in the right, arguments that amount to little more than mutual name-calling, the adoption of these extremely blurred words as the core elements of one's identity, discussions in which nobody has the time or the inclination to define terms, look for substance, or seek compromise. I myself hold what is perhaps a rather idealistic view – that a good portion of the ruling elites, and of people in general, genuinely want their fellow citizens to be happy, but have very different plans for how to achieve that. Day to day, though, I cannot escape the narrative that one social group wants to destroy another – that my life is threatened by immigrants, the ruling party, the opposition, people of a certain skin colour, or the followers of a particular religion. Above all, I find the people who have managed to attain some measure of power to be far too pragmatic, and I find it hard to believe they would spend their time and resources just to make my life worse. I prefer to believe that they would simply like to be happy – together with me. But how are they supposed to convey that to others, using nothing but words that everyone is free to interpret in their own way? On top of this comes the regular, systematic shifting of the meanings of terms used in political discourse, practised by the media. Language has become – for elites, for the media, but also for "consumers," for Internet users – a battlefield. In a world with an ideal language, or with the direct transmission of mental states, would we have a better chance at peace – and survival? Perhaps it is in the deep structures, or in what Wittgenstein calls the mystical, that the common human traits lie hidden which would let us understand that, in the end, more unites us than divides us. In this sense, too, the boundaries set by language may be separating us from another world – the one found in John Lennon's Imagine. In recent years we have been able to observe a new, far more dangerous phenomenon.
Metropolis sits on the water's edge and waits for its predator to arrive.
The "Dead Internet Theory" is a cross between internet folklore and solipsism. It holds that the vast majority of content ostensibly created by humans online is in fact generated by artificial intelligence networks and special programs ("bots"). It arose in response to the changes that online communities have undergone in recent years as a result of the popularisation and monopolisation of social media. Adherents of the theory often stress the feeling that, compared with the earlier, lively, colourful, diverse Internet, today's Internet feels like a great balloon that is "empty" and "sterile" – we are assailed by enormous quantities of content, yet it is all very similar to itself and devoid of humanity. This is supposed to be evidence that the Internet has been taken over by programs impersonating users, mass-producing and distributing content – largely for one another's benefit – in order to promote products and ideas.
For now, the Dead Internet Theory seems to be little more than an obscure curiosity, the product of nostalgia and shifting culture. But the considerable influence of bots on the shape of present-day society – and therefore of language – is beyond doubt. A report by researchers from the Computational Propaganda Project at the University of Oxford describes the activities of a company allegedly responsible for building a network of hundreds of thousands of fake accounts used in Polish politics and election campaigns. According to the report, roughly 45% of activity in the Russian-language part of Twitter comes from highly automated accounts. It also mentions a network of 500 bots used during the 2016 U.S. election, allegedly responsible for 2 million tweets, 400,000 of them about Donald Trump, and the enormous role such programs play there – over 60% of Americans rely on social media for political discussion. The potential of bots was further underscored by recent events on the American stock market, the so-called "GameStop affair," during which bots were allegedly used to encourage users of online forums to "pump" particular stocks. This suggests that the influence of artificial intelligence on society is becoming a matter of interest to further, highly influential groups.
This, of course, does not confirm the Dead Internet Theory – but it can certainly fuel it. Enthusiasts of internet folklore and researchers alike agree that the 2016 U.S. presidential election was a turning point for the shape of online communities. People speak of a "post-2016 Internet" – that was the first time social media showed its potential for shaping the world's geopolitical situation, and in certain corners of the Internet, talk grew louder of so-called "meme magic": changing reality through rapidly spreading, humorous images. The power of bots to shape society in Poland, too, is hard to overstate, given the role social media played in, for example, the Women's Strike protests, the Covid-19 pandemic, or the presidential election held during lockdown, with campaigns largely confined to the Internet. Their goal may be not so much to persuade others as to polarise society – for instance, in order to destabilize another state, or to draw public attention to particular issues; the media do not tell us what to think, but what to think about. Add to this the fact that social media, using secretly maintained algorithms to "match" content to a given user's preferences, can be compared not to an open space but to a network of closed rooms – rooms we cannot escape, and whose existence we are often not even aware of.
In the fight against spam, effective methods were developed for combating messages mass-duplicated in unchanged form – but thanks to advances in artificial intelligence, systems based on neural networks are learning to simulate human language ever more convincingly. As Alan Turing predicted in 1950, humanity has managed to teach the computer how to deceive us. In the murky activity of bots we can see clearly how control over the world is slowly slipping from our hands. Artificial intelligence has not yet killed the Internet – but it has changed and politicised it enormously, and hordes of artificial users, radicalising one another, deciding what to buy and what to promote based solely on the posts of other bots, is not such a distant vision. In 2019, Facebook was removing, on average, nearly 2 billion fake accounts per quarter. It is no accident that I focus here on the events of the last few years – if we believe that language is on the verge of a revolution, then the circumstances immediately preceding it will have the greatest influence on how it unfolds. Ever more groups, sensing growing advantage in doing so, are sending their armies of bots into the Internet, which in turn erodes trust and makes it possible to accuse anyone of being a "bot," quickly dehumanising those who hold different views. When, in the dark, we want to find another person, we resort to language, asking: is anyone there? Perhaps a language is possible in which, as once was the case, only another human being could answer such a question in the affirmative.
Using language in its unchanged form carries other, more remote dangers as well. So-called nuclear semiotics deals with the problem of warning people living thousands of years in the future about the dangers associated with nuclear waste storage sites. After all, protective barriers, ominous structures, and written warnings did not stop modern researchers from breaking into Egyptian tombs. It is difficult to step outside the frame of our own, culturally specific, understanding of signs and create a message that will effectively discourage people in the distant future from digging up uranium that is still radioactive – even simple comic strips can be read in different directions, yielding completely different messages, and a skull that looks menacing to us might, in the future, become a symbol of life or knowledge. A more abstract version of the same problem is the challenge of communicating with any extraterrestrial beings that might exist – humanity has already tried sending various kinds of messages into space, yet we do not know whether we would even manage to convey to another civilization the simple fact that we think. Breaking through the boundaries of language is therefore becoming not merely a philosopher's dream, but, increasingly, a necessity tied to technological progress.
Paradoxically, language is also a great torment for linguists themselves. Eco compares studying language by means of language to interlacing two hands in so intricate a way that, once you touch one finger, it becomes hard to say which hand it belongs to. Formal linguistics offers mathematical methods for studying language – but mathematics, too, remains an imperfect language, and it fails to resolve, among other things, many problems of unwanted polysemy. Nor does it answer the question of exactly what, for example in a philosophical context, we ought to be studying within language. This is the trap Wittgenstein describes as "closely examining the shape of a chess piece in order to better understand the rules of chess." He also compares the imprecision of linguistic methods for studying the world to trying to mend a spider's web with one's fingers. This recalls another comparison, used to describe Heisenberg's uncertainty principle – that studying the precise behaviour of electrons is like trying to describe a china display exactly by throwing a basketball around it and checking how it bounces off.
The revolutionary discoveries of quantum mechanics gave us the ability to describe the behaviour of even the smallest particles, changed the way we view the world, and made possible the invention of the laser, the GPS system, and the transistor. The twentieth century brought changes to physics's most fundamental paradigms – can we expect a similar revolution in the use of language? In 1952, Lewis Mumford compared the degree of complexity of language's symbolic structure to the complexity of the factories of his day. Since then, humanity has built the Large Hadron Collider and the International Space Station – incomparably more sophisticated than a 1950s car factory. Yet the basic mechanisms by which language operates remain exactly the same. The twentieth century did see the so-called "linguistic turn" – but it had no effect on how we use language day to day; if anything, it only further highlighted its limitations. As awareness of language's imperfection grows, so does the need for a genuine breakthrough. Perhaps the cold war of language now unfolding online between bots and humans, the media war over the meaning of political terms described by Chomsky, progress in other fields, and new possibilities for processing natural language will, in the near future, bring us a revolution on the scale of the quantum one.
My propositions are elucidatory in this way: he who understands me finally recognises them as senseless, when he has climbed out through them, on them, over them.
The pull of the unknown seems to be one of the inescapable traits of being human. We see it, among other places, in the English proverb the grass is always greener on the other side, or in the motif of the forbidden fruit, so important in our culture. Seeing a wall, we wonder what stands behind it – and in discussing the shape or size of the universe, we cannot rid ourselves of the question: so what lies beyond that? It seems to me that the quotation this essay is built around – the limits of my language are the limits of my world – is, in a way, Wittgenstein playing on this human trait, inviting the reader, the human being, the philosopher, to ask this very human question: so what lies beyond? After all, he himself cannot state outright what language is unable to capture. Wittgenstein calls his own propositions a ladder. If I were to extend this metaphor further, I picture a person shut inside a house who, climbing the ladder, glimpses a window onto the world – that ideal, mystical world that slips through the frame of language. But the window is completely covered in dust, and at first nothing can be seen through it. Only thanks to recent advances in linguistics, cybernetics, neurolinguistics, neurotechnology, and cognitive science have we managed to wipe some of that dust away and catch a glimpse of what lies beyond it – something humanity could, for millennia, only guess at. As my own attempt to cast the ladder aside and focus on the world beyond the boundaries of language, I will try to briefly describe three ways in which we might manage to touch that world, and what the effects of doing so might be.
Technology offers practical ways of pushing against the boundaries of language. When Sapir wrote that the human mind cannot be observed directly, cannot be measured, photographed, or recorded on tape, cybernetics was not yet so advanced. While we still cannot observe the mind directly, we can now observe the brain. The American agency DARPA (the Defense Advanced Research Projects Agency) funded, among other things, the NIA project (Neurotechnology for Intelligence Analysts), aimed at detecting threats visible on camera feeds more quickly. The operator, wearing a special helmet that reads brain waves, watches images that change very rapidly. At the same time, an electroencephalograph monitors the operator's brain for the P300 wave – a signal the brain produces when it notices something important. This wave appears as little as 300 milliseconds after the stimulus is presented. Thanks to this cooperation between computer and human brain, the system is able to detect significant events on the battlefield far more accurately and quickly than before.
This process seems to skip over the stage of transforming thought into natural language. Is the P300 wave already a sign? If a particular pattern of brain activity is the signifiant here, and the fact that something important is happening is the signifié, then the influence of one's cultural background is minimal here, if present at all. The sheer speed of these processes – the operator's brain processes 10 images per second – suggests there is no time here for language, that the whole thing happens unconsciously. The very fact that, once we set language aside, the human brain is capable of processing information at such speed, and more accurately than a computer, together with how effective a system built on the cooperation of the subconscious and the computer turns out to be, suggests considerable potential in this kind of approach. Here we touch on linguistic universals: the P300 wave is detected in test subjects regardless of language. Even though many applications have been found for it since it was first described, its meaning remains fairly indeterminate – only that the subject's mind has detected some anomaly. If we treat the brain waves triggered by different kinds of events as their signs (an index, in Peirce's sense), then P300 may be the first letter of a much larger alphabet of brain activity. Neurophysiology today knows many more such markers, and far more effective ways of studying them. I see in this the first steps toward the possibility, mentioned earlier, of "showing" others our feelings with less risk of misinterpretation – if we were to master the ability to record a mental state without loss, and then reproduce it in another person's brain. At the same time, as the results of the NIA project show, operating on signs like P300 is much faster than communication in natural language, and in theory more precise and closer to the deep structures – and therefore closer to the mystical as well. Perhaps working out a full "alphabet" of such waves – models of the brain activity responsible for different kinds of fear, happiness, love – would lead Wittgenstein to conclude that it is no longer the limits of language but the limits of the mind that are the limits of one's world. And that alone would be an extraordinary advance, and, it would seem, a far larger world. Recent years have brought enormous progress in brain-computer interfaces. It is already possible, among other things, to write characters purely in one's thoughts and have the data collected by an EEG during this process converted into sentences. It is said that whoever reads books lives twice. Science today knows how to transmit even terabytes of data per second – how rich a life might we lead if we could "read" a book, along with all its commentaries, in the blink of an eye? We could spend the time saved thinking carefully about its message, not to mention the possibility of becoming acquainted with every philosophical work available in the world – all in a single afternoon.
In recent years there was also another breakthrough – the neural-network-based, "deep learning" text-generating models already mentioned in connection with bots. In 2019, OpenAI released GPT-2 (Generative Pre-trained Transformer 2), a 1.5-billion-parameter system that synthesises the next object in a sequence (the next word in a sentence). It can write texts and summaries, and answer questions. GPT-2 was not initially available to everyone – its creators cited the excessive risks associated with it, for instance around disinformation. A year later its successor, GPT-3, was developed – more than a hundred times more powerful, and handling with ease many questions the previous model had gotten wrong. GPT-3's source code was not released, but one can get a sense of its capabilities through the program AI Dungeon, where it is used to generate an endless text adventure involving the user, modelled on imagination-driven games like Dungeons & Dragons.
The geometric acceleration of this technology's development means that many aspects of it remain unclear – there is no definitive answer to whether GPT-3 passes the Turing test, what the consequences of its spread might be, or how to control it. Many people wonder about the limitations of such models – are they capable of writing literary fiction, of surprising us, of forcing us to reflect? In my own opinion, after a brief acquaintance with the capabilities of GPT-3 and GPT-2, the answer to all three questions is yes. Poetry (unrhymed, in particular) seems to be a surprisingly strong suit of theirs, thanks to the sheer multiplicity of interpretations even a seemingly meaningless message can support. There is enormous potential here for further research in ethics, literary studies, translation studies, psychology, and many other fields. Perhaps such models – which, in the course of their linguistic "training," come into contact with more text than any single human being ever could – will allow us to answer some very basic questions about the nature of creativity or aesthetics. In what ways might they accelerate the evolution of language?
First, they are an excellent motivator for it. On the one hand, the inability to distinguish text written by a human from text generated by a computer carries, as the activity of bots demonstrates, an enormous threat to peace, and therefore to humanity as a species. If humans do not want to lose their position in the world, they must make machines start losing the "imitation game" again. On the other hand, such models hand linguists an enormous array of new possibilities – practically unlimited amounts of text to study, a new kind of dynamically changing corpus, insight into the mechanisms of text processing. And, at least on the surface, these natural-language models have a great deal in common with the human brain:
- They learn language on the basis of "innate" foundations, followed by large-scale contact with language.
- They function thanks to small electronic units that transmit electrical signals (in both cases, these are called neurons).
- The details of the processes occurring between the intake of new information and the generation of language remain unclear in the case of language, and are becoming increasingly unclear in the case of neural networks.
The creators of GPT-3 admit that one of the system's weaknesses is that it is hard to interpret its decisions about which word to choose next. The model is exposed to gigantic quantities of text, and its behaviour is governed by hundreds of billions of parameters – so it is hard to describe exactly what is happening inside it. In this respect, neural networks are starting to become as much of an "enigma" as the human mind. Alan Turing foresaw this too, 70 years ago: An important feature of a learning machine is that its teacher will often be very largely ignorant of quite what is going on inside. The human brain, however, has been a mystery from the moment we first began to study it – the evolution of the machine, by contrast, is very well documented, from the relatively simple "Manchester Mark I" to the at-times downright frightening GPT-3.
These similarities, together with the sheer volume and specific character of the scientific material supplied by natural-language-processing models, encourage the belief that they will help us better understand human beings, their brains, and their language. There is a risk, though, that instead of studying humans we will end up studying a program – and even a minor difference in their underlying mechanisms could lead to significantly different results. Perhaps, even so, such an approach will let us formulate useful (if imperfect) scientific theories. Lewis Mumford points to another danger arising from the human-machine relationship, writing that "the humanization of the machine may have the paradoxical effect of mechanizing the human being." I see in this yet another motivation for humanity to pursue a linguistic revolution – the desire to prove our superiority over the computer. The kind of complex that GPT-3's abilities can provoke is visible in the words of one Twitter user: We aren't pulling the mask off the machine to reveal a genius wizard, we're pulling the mask off each other to reveal the bar is low.
Another way of bending the frame of language is offered by asemic writing – a phenomenon inspired by handwriting, sitting on the border between literature and art, that consists of arranging texts out of marks which resemble letters but are not, in fact, letters. It is a hobby popular among, among others, the creators of constructed languages and calligraphy enthusiasts. The resulting works look like fragments of writing systems, evoke associations in the "reader," yet carry no predetermined meaning whatsoever. The existence, in Chinese writing among other systems, of pictograms (for instance, the character resembling a tree, 木) and ideograms (up, 上, and down, 下) suggests that some signs can be understood without prior instruction – whether, as with pictograms, through their resemblance to what Peirce calls the object, or, in the case of ideograms, through a combination of associations. Within Peirce's triadic classification, signs of this kind seem to function more as symbols than as icons, and are therefore closer to what they represent. Could there, then, exist a writing system based solely on connections between sign and object that are understandable to most people – a system whose meaning would depend less on the recipient's cultural background, and would therefore be less negotiable? Ideally, this would be a writing system understood by everyone, capable of conveying even abstract ideas to people from a completely different time and place. Ideograms and pictograms, however, make up only a small part of the Chinese writing system, refined over many thousands of years. It is also hard to believe that the creator of such a system could fully detach themselves from their own cultural background. This is where I see a possibility for asemic writing – perhaps such a system, like many things that once seemed unthinkable, could be invented "by accident"? Perhaps one day an asemic work will emerge that, despite its apparent lack of words, conveys the same idea to people on every continent, of every age. On its basis, one could then build a writing system, and after that, an ideal language – the kind of language mentioned by both Russell and Wittgenstein.
I admit that inventing an ideal language "by accident" is highly unlikely. Yet it cannot be denied that asemic writing pushes against the boundaries of language as we usually understand it. On internet forums devoted to this phenomenon, alongside their own works, enthusiasts also share "finds" – patterns on birch bark, on moth wings, or in the specific arrangement of leaves. Some of these genuinely look like sentences, poems, or words, complete with their own systems of punctuation and rhythm. From this one could draw a very optimistic conclusion – that the world is sending us messages and writing sentences on every available material. Perhaps these sentences really are meaningless, or perhaps we simply cannot understand them yet. This beautiful idea of "reading" the world is not new – according to certain traditions, God communicated with Adam through lightning and thunder, and the original Torah was written in black fire on white fire. Some linguists put forward the idea that language allows us to broaden our horizons – that it is thanks to language that a person who has seen only a single elephant in their life is able to speak of herds of elephants. So what might we be able to talk about if the whole world were our alphabet? Might humanity, in the distant future, be able to communicate anger through lightning, sadness through rain, love through the stars? This is an idea far more remote than neural networks or Neuralink – but an exceptionally inspiring one.
A find of asemic writing: a black arches moth (Lymantria monacha). Photo: Kurt Kulac, CC BY-SA 3.0, via Wikimedia Commons.
These are the inspirations that guide my thinking about the future of language. Presumably the coming changes will not touch everyone – just as physics still has not reconciled classical and quantum mechanics, and there is still room for handicraft in an age of mass production. The purpose of language is communication, just as the purpose of walking is getting from one place to another. Imagine two children who have both learned to walk; one of them goes on to learn to run, the other to dance. I see the possibility of a similar dichotomy in the future of language – on one side, a "running" language: fast and precise, closer to the language of machines, capable of processing enormous quantities of data; on the other, the "human" one – a "dancing" language: deep, and what distinguishes a human being from a robot. Predicting the exact differences between the two leaves enormous room for further research. My thesis is that these differences will be vast – deeper than the differences between individual natural languages – and that we will observe them first within online communities. Already we can see enormous differences between the use of language on Twitter, which limits posts to 280 characters and is, to a large extent, dominated by bots, and on discussion forums that place a premium on long, original posts. At first glance, it is the most popular platforms that seem to shape society to the greatest extent. In many cases, however, it turns out that the ideas that spread across the biggest social media platforms actually originate on a handful of relatively small, obscure forums. This was the case, for instance, with the word "julka," or with a very popular image of a sad green frog. This trend will likely only grow stronger – a small number of creative, deep communities will produce content which is then amplified and lightly edited by bots before reaching the mainstream. This is directly tied to the human capacity for the creative use of symbols, a trait the "dancing" language I mentioned would rest on, and which the "running" language would discard. This dichotomy alone could be a very effective driver of language's evolution, and the blending of these two modes of communication might allow us to find the long-dreamed-of, ideal way of communicating – just as a theory uniting the premises of classical mechanics, built on certainties, and quantum mechanics, built on probabilities, would let us formulate a "theory of everything."
A role similar to that of the quotation about the limits of language – an invitation to think beyond earlier patterns of thought – has recently been played, for me, by the term transhumanism: the belief that homo sapiens is not the final stage of human development, and that we will be able to reach the next stages of our evolution in a far more controlled way, through technology. One element of this evolution, to my mind, is precisely the overturning of the limits of language. Just as, according to transhumanists, the development of the human species will, thanks to technology, become clearly directed and will cease to depend on random mutation and natural selection, so too might language enter a stage of more controlled evolution – one no longer based on murky neologisms, the blurring of concepts' meanings, and the spread of particular metaphors. I do not mean here some extreme form of prescriptivism, or institutions that police language along the lines of the Académie française, but rather an evolutionary leap: the discovery of a system that would let us communicate without misunderstanding, speak of what lies beyond the world, and transmit data between minds at the speed we can observe today in computers. Science today knows ways of transmitting terabytes of data within a single second. Perhaps achieving effects like these requires shedding the slowing influence of natural language altogether. Chomsky argues that language is the trait that sets the human species apart from animals. At the same time, Elon Musk, who has already managed to change our world on many fronts, promises to make language obsolete within the next decade or so. It is hard not to notice that, as technology advances, we are drawing ever closer to the limits of language. Perhaps overturning them would allow us to become something more – and to build a new, more wonderful world.