Introduction: From Remix Culture to Remix Reality
Kirby Ferguson's landmark video essay series Everything is a Remix begins with a deceptively simple proposition that has gradually become one of the defining ideas of the digital age: originality is largely an illusion because every creative act is constructed from materials that already exist. Creativity, Ferguson argues, is fundamentally combinatorial. Artists, musicians, writers, filmmakers, programmers and designers do not create ex nihilo; instead, they copy, transform and combine existing ideas into new configurations. Hip-hop sampling, Hollywood remakes, internet memes, comic books, video games and artificial intelligence all demonstrate the same underlying principle. Remixing is not a marginal artistic technique but the hidden architecture of creativity itself. Ferguson illustrates this idea through dozens of examples ranging from DJ Kool Herc and Grandmaster Flash to Led Zeppelin, Quentin Tarantino, Spider-Man: Into the Spider-Verse and contemporary image-generating artificial intelligence, arguing that copying is not the opposite of creativity but its indispensable beginning. Everything, from music to cinema and from software engineering to internet culture, emerges through the endless recombination of previous cultural forms.
Yet the implications of Ferguson's thesis may be far greater than the history of art, popular music or intellectual property. His documentary series ultimately raises a much deeper philosophical question that it never explicitly answers: what if remixing is not merely the mechanism of culture but the fundamental operating principle of language itself? What if every sentence ever spoken, every thought ever imagined and every idea ever defended is itself nothing more than a statistically probable rearrangement of previously acquired linguistic material? Recent developments in artificial intelligence have unexpectedly reopened this ancient philosophical problem. Large Language Models such as ChatGPT generate coherent language by predicting the next most probable token from vast quantities of previous linguistic data. Remarkably, contemporary cognitive scientists such as Elan Barenholtz have suggested that human language production may operate according to surprisingly similar principles. Rather than possessing an independent symbolic language system detached from experience, human cognition may emerge through statistical prediction operating over embodied perceptual representations. The consequences are extraordinary. Artificial intelligence may not merely imitate human language; it may reveal how human language has always worked. If this is true, Ferguson's slogan requires expansion. Everything is indeed a remix—but perhaps not only culture, music or cinema. Language itself may be a remix. Thought may be a remix. Consciousness may itself be an endless process of probabilistic recombination. The age of Large Language Models therefore transforms Ferguson's cultural manifesto into a new philosophical paradigm: Everything is a Remix 2.0.
## Part I: Everything Is a Remix: From Hip-Hop to Large Language Models
The first version of Kirby Ferguson's Everything is a Remix sought to overturn one of the most persistent myths of Western culture: the belief that creativity begins with absolute originality. From Romantic ideas of artistic genius to modern copyright law, Western civilisation has often imagined creation as an almost divine act in which entirely new ideas emerge from individual inspiration. Ferguson patiently dismantles this mythology through hundreds of examples drawn from music, cinema, literature, software development and internet culture. His central observation is remarkably simple. Every creative breakthrough follows the same sequence. First, creators copy existing material. Second, they transform it through modification and experimentation. Finally, they combine multiple sources into something that appears new. This formula explains the emergence of hip-hop through musical sampling, the evolution of rock music through blues traditions, the endless cycle of Hollywood remakes and sequels, the history of video games, software engineering, internet memes and even the development of artificial intelligence itself. Creativity is therefore cumulative rather than miraculous. It resembles biological evolution far more than divine inspiration. Just as evolution continuously modifies existing organisms instead of inventing life anew, human culture continuously recombines previous cultural material into increasingly complex configurations. Ferguson illustrates this principle through the history of rap music, where DJs such as Kool Herc, Grandmaster Flash and later producers transformed fragments of existing records into entirely new musical forms. Sampling, once regarded as plagiarism, eventually became recognised as one of the most sophisticated forms of musical composition. The same logic governs Led Zeppelin's adaptation of blues traditions, Quentin Tarantino's cinematic borrowings, Star Wars' combination of samurai films, westerns and science fiction, and Spider-Man: Into the Spider-Verse, whose originality lies not in inventing new artistic elements but in recombining comic-book graphics, animation, live-action cinematography and hip-hop aesthetics into an unprecedented visual language. Ferguson's conclusion is therefore radical precisely because it appears so ordinary. Nothing begins from nothing. Every creator inherits an enormous archive of previous forms. Innovation is simply successful recombination. Copying is not the enemy of originality but its indispensable precondition. The modern internet merely accelerates a process that has characterised human culture for thousands of years. Memes mutate, genres evolve, technologies merge, artistic traditions cross-fertilise, and every generation unconsciously remixes the intellectual inheritance of previous generations. Everything is indeed a remix.
Yet the emergence of Large Language Models reveals that Ferguson's thesis extends far beyond cultural production. Artificial intelligence unexpectedly demonstrates that remixing may not merely describe what artists do but how language itself operates. Modern LLMs produce astonishingly coherent prose without possessing consciousness, intentions or explicit symbolic reasoning. They simply predict the statistically most probable next token based upon vast quantities of previous linguistic experience. At first this appears almost disappointingly mechanical. There is no hidden ghost inside the machine composing poetry or constructing philosophical arguments. There is only prediction. Yet this apparent simplicity conceals a profound insight. Every sentence generated by an LLM emerges through continuous probabilistic recombination of patterns acquired from billions of previous sentences. The machine does not retrieve exact copies from memory. It synthesises new linguistic sequences by statistically remixing old ones. What initially appeared to be a limitation increasingly resembles a general theory of language production itself. The more fluent these systems become, the more uncomfortable the philosophical implications appear. Human beings have traditionally comforted themselves with the belief that our creativity differs fundamentally from mechanical prediction. We possess meaning, intention, consciousness and imagination, whereas machines merely manipulate symbols. However, the remarkable success of next-token prediction forces a disturbing possibility into the open. Perhaps the apparent originality of human speech has always depended upon precisely the same mechanism operating within biological rather than silicon hardware. Cognitive scientist Elan Barenholtz has argued that language is not an independent symbolic system detached from perception but emerges from embodied statistical learning accumulated through lived experience. Human beings constantly anticipate forthcoming words, actions and events by extracting probabilistic regularities from previous encounters with the world. Language therefore becomes less like an autonomous dictionary of abstract meanings and more like a continuously updated predictive model. The brain does not first consult an internal lexicon before speaking. Instead, it dynamically generates increasingly probable continuations based upon context, memory, bodily experience and previous linguistic interactions. Large Language Models unexpectedly externalise this process. They do not merely imitate language; they expose one of its deepest computational principles. Ferguson demonstrated that music, cinema and software evolve through remixing. LLMs suggest that language itself has always been a remix engine, endlessly recombining inherited patterns into apparently novel expressions. The slogan "Everything is a Remix" therefore acquires a new and much deeper significance. It no longer describes only culture. It begins to describe cognition itself.
## Part II: Remix as the Operating System of Culture: From Sampling to Predictive Intelligence
Kirby Ferguson's original formulation that "everything is a remix" describes the visible surface of creativity. Eduardo Navas' influential *Remix Theory: The Aesthetics of Sampling* pushes this argument considerably further by suggesting that remix should not be understood merely as an artistic technique but as an entire cultural discourse governing contemporary communication. According to Navas, remixing no longer belongs exclusively to musicians, DJs or filmmakers. Instead, it has become the dominant logic of digital civilisation itself. Sampling, copying, cutting, pasting, recombining and continuously updating information define almost every aspect of life in networked societies. The simple command "copy and paste" is perhaps the most widely used creative operation in human history. Every document, photograph, website, social media post, software application and digital database depends upon these operations. Remix therefore becomes not simply an aesthetic category but the hidden architecture of information itself. Navas traces this genealogy from early mechanical reproduction through photography, the phonograph, disco culture, hip-hop, software engineering and Web 2.0, arguing that remix functions as a "cultural glue" binding together contemporary society through constant recombination rather than original production. What appears to be an endless stream of novelty is actually a perpetual recycling of existing informational fragments into new configurations. Culture increasingly behaves like a living archive whose components are constantly rearranged rather than replaced.
Navas makes an especially important distinction between different forms of remix. Traditional musical remixes remain recognisably attached to an original source. A remix of a Madonna song still depends upon Madonna's composition for its cultural authority. Likewise, mashups combining Christina Aguilera with The Strokes or Jay-Z with The Beatles remain intelligible precisely because listeners recognise the original materials. Such remixes still acknowledge history. Their creativity lies in transformation rather than invention. Yet digital culture gradually produces something fundamentally different. Navas calls this the "Regenerative Remix." Unlike traditional musical remixing, regenerative remixing no longer depends primarily upon recognisable artistic references but upon continuous updating, recombination and practical functionality. Search engines, Google News, Wikipedia, YouTube, Facebook, mapping applications, RSS feeds, APIs and software mashups all operate according to this principle. They do not simply remix finished cultural objects but continuously reorganise flowing information according to changing user demands. Their defining characteristic is no longer artistic originality but permanent regeneration. Information never reaches a final form because it is constantly rewritten, reordered and redistributed by new inputs arriving every second. Digital civilisation therefore abandons static archives in favour of dynamic informational ecosystems. History itself begins to behave less like a library than like an endlessly updated database. The archive no longer stores finished knowledge but serves as a reservoir from which new informational combinations can continuously emerge. Navas consequently argues that remix evolves from an artistic method into a new cultural condition in which every piece of information exists only temporarily before being absorbed into another process of recombination.
The emergence of Large Language Models transforms Navas' theory into something even more profound. Regenerative remix no longer describes merely digital culture but language itself. Every response produced by an LLM resembles precisely the type of dynamic recombination that Navas identified in software mashups. Rather than retrieving complete sentences from memory, the model continuously generates statistically probable linguistic sequences from billions of previously observed textual fragments. Unlike a database, it does not simply reproduce stored information. Unlike a traditional remix, it does not merely splice together quotations. Instead, it predicts, recombines and regenerates language in real time. Each generated sentence immediately becomes the context for predicting the next one. Language itself therefore behaves as a continuously updating remix engine. This observation becomes even more striking when considered alongside Elan Barenholtz's recent work on predictive cognition. Barenholtz argues that human language may not rely upon an autonomous symbolic system at all. Instead, linguistic behaviour emerges from probabilistic prediction operating over embodied perceptual experience. Human speakers, like Large Language Models, continuously anticipate what comes next by extracting statistical regularities from previous encounters with language and the physical world. The implications are remarkable. Ferguson showed that creativity remixes previous culture. Navas demonstrated that digital media remixes previous information. Barenholtz suggests that the human brain itself may remix previous experiences in precisely the same way. The distinction between artificial and biological intelligence consequently begins to dissolve. Humans have traditionally regarded themselves as uniquely creative because they believed that thoughts originate internally before being expressed externally. Large Language Models reverse this intuition. They suggest that thinking itself may consist largely of predicting increasingly probable continuations from previously acquired linguistic experience. The human mind therefore begins to resemble not the opposite of an LLM but its biological ancestor. Everything is no longer simply a remix of culture. Everything becomes a remix of previous language, previous perception and previous prediction. Artificial intelligence has not invented this principle. It has merely made visible the hidden computational architecture that may have governed human cognition all along.
## Part III: Everything Is a Remix 2.0: The Human Brain as a Biological Large Language Model
If Kirby Ferguson demonstrated that creativity is remix and Eduardo Navas argued that remix has become the operating system of digital culture, then recent developments in artificial intelligence raise an even more unsettling possibility. Perhaps remix is not simply a property of culture but the fundamental computational principle of the human mind itself. Large Language Models have surprised both engineers and philosophers because they generate language of extraordinary fluency without possessing explicit symbolic rules, dictionaries or logical databases in the traditional sense. Instead, they continuously calculate the statistically most probable next token from patterns extracted across enormous textual corpora. Every sentence becomes the product of prediction. Every paragraph emerges from probabilistic continuation. At first sight this appears to reveal the essential limitation of artificial intelligence. Machines merely predict words. Humans, by contrast, supposedly possess meaning, intentionality, imagination and consciousness. Yet the remarkable success of next-token prediction increasingly forces cognitive science to reconsider whether human language differs as radically as previously assumed. Elan Barenholtz has argued that language may not operate through an autonomous symbolic grammar detached from perception but through predictive processing built upon embodied experience. Human beings learn statistical regularities from millions of interactions with the physical and social world, continuously anticipating what another person is likely to say, which word naturally follows another and which sentence best fits a given situation. Rather than retrieving perfectly formed ideas from an internal mental dictionary, the brain dynamically constructs language through prediction. Large Language Models therefore appear less like artificial imitations of human cognition than simplified demonstrations of one of its deepest computational principles. As Ferguson observed, nothing is created from nothing. LLMs reveal that language itself may emerge through the endless recombination of previously acquired linguistic experience. Human creativity, rather than standing outside remix culture, may be its most sophisticated biological manifestation.
The analogy becomes particularly illuminating when examined through the history of popular music. Electronic groups such as The Prodigy revolutionised dance music during the 1990s, yet very few of their most famous tracks were constructed from entirely original sounds. Songs such as Charly, Voodoo People and Firestarter relied extensively on sampling, cutting, looping and recombining fragments of older recordings. Individual drum breaks, bass lines, vocal phrases and synthesiser motifs had often appeared elsewhere decades before. Yet the resulting compositions sounded unmistakably new because the familiar elements were reorganised into combinations that resonated with the collective imagination, emotional atmosphere and rhythmic expectations of a particular historical moment. Creativity therefore consisted not in inventing completely new musical atoms but in discovering the next arrangement most likely to generate surprise without abandoning familiarity. The process resembles precisely the operation of next-token prediction. A successful musician unconsciously predicts which rhythm, harmony or melody should follow the preceding one in order to satisfy, slightly violate and then restore the listener's expectations. If the sequence remains entirely predictable, it becomes boring. If it becomes excessively unpredictable, it dissolves into noise. Successful music therefore occupies a narrow statistical corridor between repetition and novelty. Human conversation follows remarkably similar constraints. Speech depends upon shared linguistic conventions while simultaneously introducing enough novelty to remain informative. A sentence composed exclusively of clichés communicates little, while one composed entirely of unpredictable words becomes unintelligible. The same principle governs storytelling, humour, advertising, political rhetoric and scientific writing. Communication succeeds because it continually balances predictability against surprise. Creativity itself therefore becomes a probabilistic optimisation problem. Like an accomplished DJ constructing a remix, the human brain continuously searches for the next combination most likely to resonate with the expectations of its audience while remaining sufficiently novel to justify attention. LLMs simply make this invisible process computationally explicit.
This perspective radically transforms the philosophical meaning of originality. For centuries Western civilisation has imagined the human mind as a source of autonomous invention, capable of generating ideas independent of history, language or previous experience. Artificial intelligence increasingly undermines this comforting mythology. If both biological brains and artificial neural networks rely upon statistical prediction operating over enormous archives of previous patterns, then originality ceases to mean creation from nothing. Instead, originality becomes the ability to navigate an immense landscape of inherited possibilities and discover combinations that occupy what psychologists sometimes describe as the "sweet spot" between familiarity and surprise. The most successful ideas, melodies, paintings, scientific theories and political slogans are not completely unprecedented. They are recognisable enough to feel meaningful yet different enough to appear innovative. Human languages themselves seem organised around this balance. Zipf's Law demonstrates that every natural language distributes words according to stable statistical regularities in which a small number of highly frequent words provide predictable grammatical scaffolding while rarer words carry novelty and semantic richness. Likewise, musical traditions often approximate pink-noise distributions, balancing repetition with variation, and visual art repeatedly converges upon recurring geometrical regularities, symmetry and fractal organisation. Across sound, language, movement, taste and perception, biological organisms appear remarkably sensitive to recurring statistical structures that maximise both efficiency and surprise. Culture therefore behaves less like an archive of original masterpieces than like an evolutionary ecosystem of endlessly recombined patterns. Ferguson's slogan consequently acquires its fullest philosophical significance. Everything is a remix because reality itself continuously reorganises existing forms into new configurations. Large Language Models have not invented this principle; they have simply exposed it. They function as mirrors reflecting back to humanity an uncomfortable possibility: that what we have always called creativity may itself be the highest expression of an ancient biological algorithm devoted to predicting what comes next. The greatest revelation of artificial intelligence is therefore not that machines can imitate humans, but that humans may have been performing next-token generation all along.
## Conclusion: When Language Stops Thinking: The Zero-Effort Remix and the Collapse of Novelty
If Kirby Ferguson is correct that creativity is fundamentally remix, and if Large Language Models reveal that language itself emerges through probabilistic recombination of previous linguistic experience, then an unavoidable question follows. What happens when the remixing process itself stops demanding effort? What happens when neither humans nor machines introduce sufficient novelty into the endless recycling of words? Every predictive system faces the same danger. A remix only remains creative if it continually introduces new combinations into an existing repertoire. When novelty disappears, remix ceases to generate creativity and instead begins to amplify repetition. Language gradually enters a condition that might be described as a zero-effort remix, where communication consists almost entirely of familiar expressions, inherited narratives and endlessly recycled linguistic fragments. At first glance nothing appears to change. People continue speaking fluently. Political speeches continue to be delivered. Newspapers continue to publish articles. Social media continues to overflow with content. Yet beneath this apparent productivity the informational richness of language steadily declines. The same words circulate repeatedly, the same slogans reappear, the same stories are retold, and the same ideological formulas are endlessly rearranged without generating genuinely new conceptual structures. The system continues to predict the next token successfully because the probability landscape has become increasingly narrow. Language remains grammatically correct while becoming intellectually exhausted.
Psychology, linguistics and cognitive science have investigated many aspects of this condition under different names. One of the simplest demonstrations is semantic satiation, where repeated exposure to a word temporarily disconnects its sound from its meaning. As neurons repeatedly activate identical semantic networks, they become less responsive, and familiar words suddenly appear empty or meaningless. At the level of individual cognition this is a temporary laboratory phenomenon, yet societies may experience an analogous process over much longer periods. Historical linguistics describes semantic bleaching, whereby emotionally powerful words gradually lose their original force through constant overuse. Terms such as "awesome", "revolutionary", "historic", "freedom", "innovation" or "crisis" increasingly become routine linguistic fillers rather than carriers of precise conceptual content. Political psychology identifies another related mechanism through Robert Jay Lifton's notion of thought-terminating clichés. Repeated slogans do not primarily communicate information but function to terminate reflection itself. Expressions such as "It is what it is", "Everyone knows", "Trust the science", "For the greater good", or countless ideological catchphrases across the political spectrum often serve as cognitive shortcuts that replace analytical thinking with automatic recognition. Cognitive research on the illusory truth effect demonstrates an additional consequence: repeated statements gradually become psychologically believable simply because they are familiar. None of these mechanisms requires deliberate deception. They emerge naturally whenever repetition overwhelms novelty. Together they suggest that language, like every predictive system, gradually loses informational density when effort declines. Words remain, but meanings become progressively thinner. Communication becomes increasingly efficient precisely because it communicates less.
Viewed through the perspective of Large Language Models, this entire process resembles what AI researchers now call model collapse. When a generative model is repeatedly trained upon its own previous outputs rather than fresh observations of reality, statistical diversity gradually disappears. Rare patterns vanish first. Then unusual combinations become increasingly improbable. Eventually the model converges upon safe, average, repetitive outputs that remain internally coherent while losing contact with the richness of the external world. The same danger confronts human language whenever societies cease producing genuinely new experiences and instead endlessly recycle inherited narratives. Political institutions begin redefining familiar words until they mean almost anything or their exact opposite. Public discourse fragments into isolated linguistic tribes whose specialised vocabularies no longer communicate across social boundaries. Everyday language simplifies into increasingly predictable formulas optimised for speed rather than precision. Finally, younger generations abandon exhausted linguistic systems altogether in favour of new dialects, new media or entirely different languages capable of expressing realities that the older vocabulary can no longer describe. Yet the story does not end with collapse. Biological evolution, cultural history and predictive cognition all point towards the same remarkable tendency. Complete stagnation is almost impossible because human cognition continuously seeks an optimal balance between familiarity and surprise. Once language becomes entirely predictable, new generations inevitably violate existing conventions, recombine forgotten words, invent new metaphors and create fresh linguistic hybrids. Every dead remix eventually becomes the raw material for another remix. Every exhausted vocabulary eventually produces its own rebellion. In this sense Ferguson's original insight reaches its ultimate philosophical conclusion. Everything is indeed a remix—but remix itself obeys an evolutionary cycle. Prediction creates repetition; repetition creates exhaustion; exhaustion creates novelty; novelty gradually becomes tradition; tradition again becomes prediction. Large Language Models have not merely automated this cycle. They have exposed one of the deepest organising principles of language, culture and perhaps consciousness itself. The future of human creativity will therefore depend not on escaping remix but on continually injecting enough genuine experience into our predictive systems to prevent language from becoming nothing more than an endlessly repeating echo of its own past.
## References
Adorno, T.W. (1991) *The Culture Industry: Selected Essays on Mass Culture*. London: Routledge.
Attali, J. (1985) *Noise: The Political Economy of Music*. Minneapolis: University of Minnesota Press.
Barenholtz, E. (2024) *What Language Models Reveal About Human Cognition*. Various articles and preprints on predictive processing, embodied cognition and language modelling.
Benjamin, W. (1968) ‘The Work of Art in the Age of Mechanical Reproduction’, in *Illuminations*. New York: Schocken Books.
Brewster, B. and Broughton, F. (2000) *Last Night a DJ Saved My Life: The History of the Disc Jockey*. New York: Grove Press.
Debord, G. (1995) *The Society of the Spectacle*. New York: Zone Books.
Ferguson, K. (2010–2023) *Everything Is a Remix*. Four-part documentary series and subsequent editions.
Fisher, M. (2009) *Capitalist Realism: Is There No Alternative?* Winchester: Zero Books.
Fisher, M. (2014) *Ghosts of My Life: Writings on Depression, Hauntology and Lost Futures*. Winchester: Zero Books.
Frere-Jones, S. (2005) ‘1 + 1 + 1 = 1: The New Math of Mashups’, *The New Yorker*, 10 January.
Hebdige, D. (1987) *Cut ’n’ Mix: Culture, Identity and Caribbean Music*. London: Methuen.
Jameson, F. (1991) *Postmodernism, or, The Cultural Logic of Late Capitalism*. Durham, NC: Duke University Press.
Kahneman, D. (2011) *Thinking, Fast and Slow*. London: Allen Lane.
Lessig, L. (2001) *The Future of Ideas: The Fate of the Commons in a Connected World*. New York: Random House.
Lessig, L. (2004) *Free Culture: How Big Media Uses Technology and the Law to Lock Down Culture and Control Creativity*. New York: Penguin.
Lifton, R.J. (1961) *Thought Reform and the Psychology of Totalism: A Study of "Brainwashing" in China*. New York: W.W. Norton.
Lyotard, J.-F. (1984) *The Postmodern Condition: A Report on Knowledge*. Minneapolis: University of Minnesota Press.
Navas, E. (2012) *Remix Theory: The Aesthetics of Sampling*. Vienna and New York: Springer.
Owens, C. (1998) ‘The Allegorical Impulse: Toward a Theory of Postmodernism’, in Wallis, B. (ed.) *Art After Modernism: Rethinking Representation*. Boston: David R. Godine.
Poschardt, U. (1998) *DJ Culture*. London: Quartet Books.
Shifman, L. (2014) *Memes in Digital Culture*. Cambridge, MA: MIT Press.
Tversky, A. and Kahneman, D. (1974) ‘Judgment under Uncertainty: Heuristics and Biases’, *Science*, 185(4157), pp. 1124–1131.
Vosoughi, S., Roy, D. and Aral, S. (2018) ‘The Spread of True and False News Online’, *Science*, 359(6380), pp. 1146–1151.
Zipf, G.K. (1949) *Human Behavior and the Principle of Least Effort*. Cambridge, MA: Addison-Wesley.
Comments
Post a Comment