[1] The text took 3,000 years to put the images in museums. The current dispute will be quick and digital will be fast. Will there be a reactionary resistance of surpassed texts against computer programs in a future that cannot yet be foreseen? (FLUSSER, 2011, p. 209)
[2] Here we have in mind true writing machines (artificial intelligence) that provide their own organization [of writing]. Such machines are not, to be honest, just for writing, but also thinking machines, which should lead us to reflect on the future of writing and, in general, on thinking. (FLUSSER, 2011, p. 19)
[3] All writing is correct writing, and this indirectly causes the current crisis in writing. For there is something mechanical in organizing, in lining up, and for this, machines perform better than humans. Writing, the organization of signs, can be left to machines. (FLUSSER, 2011, p. 19)
Prompt: "The linear history flows in a channelled way, through scripts, into images, to spin, according to programming, around them in the eternal return of the same"
I was not yet born when Vilém Flusser published "Die Schrift: hat Schreiben Zukunft?" (1987), a book – also released in floppy disk format – that speculates the probable effects of artificial intelligence on writing. It was the late 1980s, but Flusser seemed to be elaborating a position from his time for the 2023 discussion on Large Language Models (LLM) such as ChatGPT.
Other books by Flusser could also serve as a basis for reflecting on the topic. The Universe of Technical Images (1983) is, in Brazil, one of the most well-known and certainly addresses the issue of artificial intelligence. A considerable part of his work, in fact, traverses this debate. However, as the discussion around artificial intelligence has intensified lately with the release of ChatGPT-4, I chose to reread and take notes with renewed attention to his book "Is There a Future for Writing?" This post will share this note-taking, offering a personal collage of excerpts to speculate on the Czech-Brazilian media philosopher's position on the future of writing after the release of GML to the general public.
Perhaps it is because long-standing issues are being updated in the existential drama of historical thinking that Vilém Flusser has played with being perspicacious and current to reflect on ChatGPT, even 30 years after his death. Several of the debates raised so far date back a long time and are discussed by him: the death of the author, the artwork without aura, the effects of automation on work, the question of consciousness, the fear that artificial intelligence will turn against humanity, etc. These are themes debated if we consider the era of mechanical reproduction in the 1930s, the era of cybernetic revolution in the 1960s, and the era of technoculture in the 1990s, all widely traversed by Flusser's studies and writings.
In the 1980s, Flusser stated that he looked at typewriters and the Word word processor with curiosity about their speed and variability, anticipating the development of their ability to reproduce correct writing:
Undoubtedly, artificial intelligence will be more intelligent in the future. It will have a historical consciousness that will far surpass our own. It will make better, faster, and more varied history than we have made. The dynamics of history will intensify towards the unimaginable: more and more things will happen, events will precipitate and become more diversified. As for us, we can confidently leave all of history to automatic machines. Since all these automatic and mechanical things will make history better than we can, we can concentrate on other things. On what? This essay is precisely about that when it asks, "Is there a future for writing?" (FLUSSER, 2011, p. 23)
I begin by drawing attention to some details in Flusser's own meta-writing (writing about writing). Firstly, I highlight when he states that texts need to be attuned and that there are two types of rhythm and textual tuning: in the first, one discourse wave follows another. In the second, the wave breaks and foams, with the text contradicting itself and yet flowing without gaps (FLUSSER, 2011, p. 74). The second would be a syncopated rhythm, and in writing this, I believe he affirms the Brazilian rhythm that underpins his thinking. He leaves an entire atmosphere of samba implicit in his vision of a technological thinking without fixed ground (bodenlos), in constant displacement between languages and imagistic, alphanumeric, and alphabetical dimensions of space-time.
Secondly, I find it interesting to highlight his assertion that texts are a search for the other (with all the ethnocentric problems embedded in the construction of alterity in Western texts). I like the image he creates that all texts are "outstretched arms that seek, with or without hope, to be embraced by another. This is the atmosphere of the gesture of writing" (FLUSSER, 2011, p. 67). With this, he asserts that texts are political, even when the writer is aware or unaware, or even indifferent, or even when "he emphatically gives up, like Kafka, on the reader as a co-producer of the text" (FLUSSER, 2011, p. 67).
These characteristics that make a text written by a human something political are not present in texts generated by AI, as they do not possess a political sense.
As an artificial intelligence language model, ChatGPT does not have a political sense or political beliefs. It does not hold opinions or take political stances. Its responses are generated based on statistical patterns and language patterns found in its training data. However, it is important to note that the questions and prompts given to ChatGPT may have political implications or biases, and the responses generated by ChatGPT may reflect these biases.
When I asked ChatGPT if it has a political sense, I got this answer:
As an artificial intelligence language model, ChatGPT does not have a political sense or political beliefs. It does not hold opinions or take political stances. Its responses are generated based on statistical patterns and language patterns found in its training data. However, it is important to note that the questions and prompts given to ChatGPT may have political implications or biases, and the responses generated by ChatGPT may reflect these biases.
Although they may be impeccably correct and fueled by the humanly unintelligible vastness of data stored in highly frigid servers in specific territories on the planet, this absence creates a risk of a "total depoliticization of all forms of behavior," resulting in humans and society being guided as a cybernetic automatic system (FLUSSER, 2011, p. 93).
In this context, Flusser believed that the alphabet will have to defend itself against being supplanted by new codes – "a consolation for those engaged in the permanence of writing texts, as things have accelerated" (FLUSSER, 2011, p. 223). The fear, therefore, is that perception and experience models generated by devices will be adopted uncritically, and that artificial intelligence will robotize life.
Furthermore, GML tends to reduce thinking and intelligence to a biased understanding limited by what thinking and intelligence mean in the West (within the current stage of scientific knowledge). In other words, we will be basing our thinking and view of intelligence on a simplified and biased projection (especially by the white Eurocentric and heteronormative power matrix) of what it can be:
In the brain, these representations are formed from distinct elements, and from them, thoughts, desires, decisions, and feelings emerge again (in a quantum way). One cannot understand how this happens in its details in view of the complexity of the brain, which borders on incredulity. But this can be simulated in a simplified way in intelligent machines.
From the inscription of symbols on rocky surfaces to writing by artificial intelligence, there is a process of thousands of years. Writing, throughout this time, has always been "a gesture that guides and aligns thought." It is a gesture that involves organizing thought into a linear structure that, in the West, was fundamental to the affirmation of logical thinking over mythical, circular thinking.
The turning point of this affirmation of progressive linearity against mythical circularity is recognized as the Enlightenment and its technology of the Gutenberg Press: "Typography was the very expression of Western, historical, scientific and progressive thought that became conscious (FLUSSER, 2011, p. 87)". Since then, and more rapidly, the West has shifted thinking from a "prehistoric" magical circle to a historical thinking conformed in lines.
"Writing, this sequence of signs in the form of lines, makes this historical consciousness possible. Only when one writes lines can one think logically, calculate, criticize, produce scientific knowledge, philosophize, and, analogously, act. Before that, one walked in circles. And the longer the lines written, the more historically one can think and act. The act of writing evidences historical consciousness, which is strengthened and deepened through continuous writing, and writing, in turn, becomes stronger and denser. This feedback between the writer and historical consciousness provides tension that intensifies and allows for ever more advancement. This is the dynamic of history". (FLUSSER, 2011, p. 21)
The writing deprived the representations by image that existed before it and the consciousness itself, giving rise to historical consciousness and its subject. The thinking of this subject rejected the zone of imaginary, magic and ritual until it reached the point of establishing an institutional place for this zone to be objectified by its one-dimensionality: the museum.
"When speaking, one speaks "about" representations by image and "about" images, one stays above imaginary thinking and speaks from top to bottom... one writes alphabetically to affirm and amplify the level of conceptual and over-imagetic consciousness, instead of succumbing continuously to plastic thinking - as in the characteristic speech of the creation of writing". (FLUSSER, 2011, p. 55)
Thus, a thinking was established that "tears our representations of the world to organize them in a ragged way into ordered lines, into concepts that can be counted, narrated and criticized" (FLUSSER, 2011, p. 31), made transparent for the contemplation of a disembodied mind.
Typography accelerated progress and "the abandonment of everything that is real for what is only possible, of all being for becoming" (FLUSSER, 2011, p. 41). The pace of this acceleration was intense for humanity and the planet: we write quickly and schematically in search of the endpoint and the future. This would justify for Flusser the asthmatic character of our writing: thinking and writing have an internal dialectic that demands taking a breath, contemplative pauses that oppose urgent impulses, breath for "critical thinking". Hurried and paused, schematic and critical are the characteristics of acting and thinking oriented in the form of a line and are the very structure of a thinking that travels a line from the past to the future, traversing the present without stopping (FLUSSER, 2011, p. 40).
Such one-dimensional thinking, however, has led to the universe of technical images where devices now play a predominant role. For Flusser, the computer revolution made typography, the alphabet, and this superfluous thinking obsolete, leading to a new mode of thinking that is now evident, but was already sensed by Flusser. Of the devices, it is highlighted that they have no existential brakes and do not need air to breathe: they do not need breath to process humanly unintelligible data at lightning speed. It is no wonder that the possibility of uninterrupted and ultra-fast progress is entrusted to them.
We are about to leave overwriting (writing itself) to the devices and focus on producing and contemplating images. We are about to move to the "universe of technical images", from where we will look with superiority at the stories automatically written by devices (FLUSSER, 2011, p. 42).
Just as foundational thinking engaged against magic and myth, code-based thinking engages against progressive ideologies to replace them with cybernetic, systems-analytical, and structural modes of thinking (FLUSSER, 2011, p. 223). Flusser believed that, in this context, the alphabet would gradually detach from spoken language through disks, tapes, and images that would inundate us from screens: something like contemporary culture around Youtube, influencers, and podcasts. In this scenario, "even artificial intelligence will learn to speak" (FLUSSER, 2011, p. 102). The possible result of this movement would involve a return of myth and a certain "barbarization" of thinking, terms that gain clear political images if we think about the Brazilian context and the use that the far-right made in 2018 of algorithms.
"Due to the unimaginable quantity of models of perception and experience that will flood society in the form of images and sounds, it is uncertain whether language, which will represent only an auxiliary code, will continue to be used to shape our perceptions and experiences. Poetic force will likely be concentrated in non-verbal codes, partially still unimaginable. This type of code will no longer be read. On the contrary, it will have to be deciphered differently. The question arises, which can no longer be avoided, about the future of reading". (FLUSSER, 2011, p. 119)
On the one hand, Flusser offers an optimistic view in which computers equipped with artificial intelligence can realize the unimaginable and expand our experiences. On the other hand, Flusser points out the need to "clench our fists" against the "growing industry of automatic anesthesia" (FLUSSER, 2021, p. 71). Relearning of thought: thought is based on images and is not a continuous discursive process: "thought quantizes" (FLUSSER, 2011, p. 219). The result of this is a thought distinct from that which characterizes Western thought, for whom thought was a progressive process that disconnects from images, representations by images, criticizing them to become increasingly conceptual. With this, he does not declare the end of alphabetic thought itself, but the need to relearn to criticize from the digital.
With this book, I believe that Flusser anticipated that writing would experience a crisis due to the advent of artificial intelligence, seeking with his syncopated meta-writing to point out that in this crisis there is a clash of temporalities and consciousness, since a dizzying acceleration of thought from the alphanumeric digital dimension corrodes historical consciousness and its subject, with the worst effect of this crisis being the corrosion of critical thought, which needs to reinvent itself in its ability to be critical through an intelligent (not artificially) creative play between images, digital codes, and the old alphabetic writing.
I asked ChatGPT for a pessimistic, optimistic, and realistic view on the effects of GML on critical thinking.
Pessimistic views of the erosion of critical thinking due to the popularization of GML (Machine Learning Language Models) may include:
Reduction in diversity of opinions and ideas: As people become more dependent on language models, they may be more likely to follow suggestions generated by algorithms, rather than seek a variety of sources and perspectives.
Increase in algorithmic bias: Since language models are trained based on existing data, they may perpetuate biases and stereotypes already present in the collected information. This can lead to algorithmic bias, which means that the suggestions generated by language models may be biased and not reflect reality.
Loss of critical thinking skills: With the convenience of easily obtaining information with language models, people may neglect the practice of analyzing information, evaluating sources, and thinking critically. This can lead to excessive dependence on algorithms and a decrease in the ability to think independently.
Spread of misinformation and fake news: With the ability of language models to generate texts similar to those written by humans, it may be more difficult to distinguish between real and fake information. This can lead to the spread of misinformation and fake news, especially if the information generated by language models is not carefully verified and evaluated.
Optimistic views on the popularization of GML and its relationship with critical thinking include:
Greater access to information: With the help of GML, people will have easier and faster access to information that was previously difficult to find or required specific skills to access. This can lead to an increase in critical thinking, as people have access to more viewpoints and perspectives on a given topic.
Development of analytical skills: The ability to analyze and evaluate information produced by GML can help develop analytical and critical skills. This can lead to greater awareness of the accuracy and reliability of the information, allowing people to better evaluate the quality of the information they receive.
Improvement in education: The popularization of GML can lead to advances in education, such as the development of machine learning programs to help teach critical thinking skills. This can help people better evaluate the quality of information and make informed and grounded decisions.
Access to new forms of thinking: GML can help create new forms of thinking and expression, allowing people to communicate and understand ideas in different ways. This can lead to an increase in creativity and innovation, as well as a greater understanding of complex concepts.
Reduction of biases and prejudices: GML has the potential to analyze large amounts of information without human biases or prejudices, which can lead to greater accuracy and objectivity. This can help reduce biases and prejudices in areas such as data analysis, decision-making, and information evaluation.
Realistic views on the popularization of GML is that it brings both opportunities and challenges to critical thinking. On the one hand, broader access to information and knowledge can stimulate curiosity, research, and the development of critical skills. On the other hand, information overload and the ease of spreading false information can also make it difficult to identify reliable sources and critically evaluate information.
In addition, the popularization of GML can lead to changes in the way knowledge is produced, disseminated, and used. For example, the development of AI tools can allow for the creation of new forms of knowledge and new research methods. On the other hand, an excessive dependence on AI tools can also lead to a reduction in human capacity for reasoning and critical judgment.
In short, a realistic view on the popularization of LLM is that it brings both opportunities and challenges to critical thinking, and that it is important to balance access to information with the ability to critically evaluate that information and make informed decisions.
Comments