Hi guys, just wanted to thank, as everyone has done, Wendy for this opportunity and everyone here, being here still on the third day after lunch. I guess that's quite good. So I'm just going to start. I have no other entertainment than myself. So you need to bear with me. I'm going to go speak slowly and read slowly. and so hopefully you can just follow. Can you hear me? Okay? Yeah, okay. Mediation today is dominated by a new order of automation. Algorithms talk to other algorithms before talking to us. They execute instructions without being supervised
and take decision without awareness. As Catherine Hales explains, explains, the automation that drives market trading, military, and commercial logistics eludes logical causality. Instrumentality and not logic, procedural operations and not postulates, we are told, determine mediatic functions today. Non-conscious automation, I argue, has radicalized the existing tension in Western culture between philosophy and technique. As technological abstraction questions
the very conditions of thinking, the bifurcation between theoretical reasoning and instrumentality falls apart. Theories of mediation on the one end reject instrumentality, but on the other, accept a media ontology grounded precisely in instrumental knowledge. We have heard a lot about this this morning and just the panel before. They've embraced imminence, as discussed by Gilles Dallais, but also Frederick Kittler's technicity of being. But whilst this theory argued for a materialist approach to mediation, the latter, I suggest,
requires a processual articulation of practical and theoretical, instrumental, and pure reasoning. I draw on Charles Sander Pierce to explain this process through what he calls a triadic architecture of logic. And I suggest that automated intelligence, as part of a computational stratum, It's irreducible to biological and physical terms. It cannot be discussed according to biology or physical sciences, but needs a science of computation itself to be discussed. A materialist account of mediation, instead, needs to address the order of this stratum,
composed of information, data, algorithms, procedures, and compressions, entropy, and randomness. These are internal constraints to computation that define how automated systems may produce knowledge, but also, and that's the challenge, develop a theoretical order. The question does is, what happens when instruments start theorizing? How does this change the critical view of instrumentality? As human thinking becomes dominated by automated activities, what conditions of mediation remain possible? But before delving in this very attractive speculation, I want to discuss first the two
main approaches that I see central to the discussion about the conditions of mediation today. So first, Deleuze. In the last 20 years, the materialist critique of representation has argued for a relational ontology, a being of the middle that has radically weakened the dominance of the semiotic and discursive understanding of mediation. In particular, I'm referring to the critique of mediation developed through the work of Gilles Deleuze and his critique of cognition specifically. of the discursive negotiation between concepts and things, Deleuze insists on a virtual field
of relations in which thought and things are equally immersed. This field is the generative source of being, or the intermezzo, as he calls it, or the self-differentiation where the sensible unfolds and forces thought to think. In difference and repetition, the less offers us a radical critique of mediation. Whilst common sense has it that every human can think because they can cognize, thinking for the less is not recognition and what has already been sized and categorized cannot
explain thinking anew. Contrary to the Kantian schema based on this ascending movement from sensible to the intelligible and finally to reasons as judgments, Deleuze's materialism proposes a radicalization of the sensible. He calls it the being of the sensible. So he ontologizes the sensible as an imminent generator of novelty that acts before and does explain cognitive reasoning. We all know that in the aftermath of poststructuralism, the adaptation of the less anti-representational view of digital media, in particular, has radicalized
this understanding of mediation. So the manifold of the sensible becomes the ontological condition that, following the Les, gels together all kinds and degrees of being. The immediacy of material reverberations and not the structure of semiotic mediation because the condition or the possibility for pure thinking. That's what the Les is about. That's what it's after in this critical cognition, pure thinking. From this standpoint, mediation also means transduction. We are all familiar with this Simondonian concept. And we know how central that has been to critical theory
of digital media. This Simondonian energetic explanation of mediation for which information becomes the manifestation of an impulse springing through the physical, biological, and technical system. As opposed to the philosophical tradition for which concepts mediate the relation between things, Deleuze offers us a practical philosophy where concepts are at one with living matter, without belonging to any particular form. For Deleuze, however, thinking is neither theoretical logic nor instrumental practice.
Closely inspired by Simon Don, and that's obviously my reading of the last Simon Don, because there are other readings with whom I don't agree, we find an immanent view of technology as matter in formation. That's the processual understanding of the last. Matter is information, and not technology as a technical operation of encoding and decoding data. There is a virtual continuum between natural and cultural strata, a meta level for the Les, okay, which exceeds for him computational logic, cybernetting mechanisms, and representational frameworks because they cannot account for something that is very important to the Les,
an infinitesimal self-differentiation, the level of imperceptible capacity of difference to differentiate itself, an internal generative concept of differentiation, which constitute what he also calls the intensity of encounters. So we know self-differentiation and intensity are very important concepts, in the list. By suspending the dominance of mathematical logic and algebraic axiomatic in favor of what he calls problematics, contingency, experimentation. Technical mediation is plunged within what can be called the biophysical dynamics.
And technoculture becomes defined in terms of natural indeterminacy. That's what we have inherited by the use of the layers in digital media, I think. But beside this energy-based critique of mediation, Frederick Kittler's media ontology has also importantly driven the debate. His theory also presents, like Deleuze, an anti-formal and practically oriented view of mediation, but does not evoke the indeterminacy of abstract materialism, as we all know. Instead, he argues for the historical production of instrumental knowledge. Kittler's theorization of media as information systems that store, transmit, and command data
develops a non-representational view of mediation for which technology is an original dimension of human culture. Not an expression, but an original dimension. The technology incorporated in media systems is neither instrument for cultural expression nor extension of sensory motor or cognitive functions. We all know how much he took, you know, a tissue, the view of McLuhan, of media, we all know that, fortunately, I mean, importantly did that. The being of technet instead is defined by the practical history of systems evolving within systems.
Hence, mediation entails the operation of information compression, data encoding, the logical procedures, the algebraic operations, defining the discrete nature of technet. And by technet, technical being, that's what it's meant. Importantly, Kittler calls for the withdrawal not of philosophy or theoretical reasoning, but of the human subject and rational judgment. He invents a materialist method in which formal logic is replaced by practical knowledge embedded in the circuits and processors of machines. This method, we all know, we have heard of Heidegger before, radicalizes the Heideggerian
view of technet. What is it? The centrality of crafting in the shaping of thought. Here technet has a messianic function, as it were, of revealing the finitude of human reason and the delusion or the delusional ends of a platonic project whose rational aim of liberating of liberating ideas from the sophist instrumentation of logos as withdrawn thinking from being altogether. But this is a longer history and we know that the distinction of technique or craft versus philosophy or rational thought strongly characterizes critical theory mistrust of technology and until today
of an automated word destined to stupidity. We have heard this this morning. In the Republic, Plato wrote of the Socratic distinction between technical or craft knowledge from philosophical logos and argued that craft knowledge only concerns technical understanding or can be used to define the pursuit of a particular trade or practice. Craft knowledge must be instrumental to something. The water clock, the astronomical array, the mechanical puppet are all primordial automated devices that are used to demonstrate or describe something.
Technet, therefore, reflects practical understanding and is distinguished from philosophical knowledge, which is rational, of a mathematical order, and requires no instruments. Whilst technique coincides with primitive automata, which are thoughtless and mainly concern practical understanding, philosophical knowledge, this is classically traditionally understood, involves the cultivation of the principle of all things, and is therefore ideal and timeless. By following Heidegger, Kittler radicalizes the transformation that technology has imposed on human culture and ultimately on thought.
If mediation is the embodiment of instrumental use and practical knowledge, the World War II computational turn marked an epochal change in which practical knowledge could finally acquire autonomy. That's why Hitler is so interesting to me. He says that very clearly and very uniquely still in media theory. With the mechanization of logic, reasoning became incorporated in computational procedures, allowing the mediatic apparatus to develop its own technical language, Whilst arguing for a historical formation of this technical ontology, Kittler reveals
the significance of instrumental knowledge as this became rooted in the information processing of the Turing machine. The distinction between theoretical and practical reasoning is here replayed in the distinction between software and hardware. The Turing machine proposed a digital uniformity between data and programs, entailing no physical difference between processor and processed. In the famous essay, There is no software, Kittler shows that the logic computation is absorbed, the logical computation is absorbed in the practical knowledge of circuits, tapes, macro possessor, and switches.
By reversing theoretical into practical thinking, that's possibly the philosophical operation that I'm trying to understand what he does. So by reversing theoretical into practical thinking, Kittler reveals that the material storage, transmission, and command of information has produced an autonomous system of knowledge derived from the automation of human activities, or quoting Wiener, from the human use of human beings. But how can these influential approaches help us to tackle the tension between automation and philosophy, i.e. the ontological autonomy of instrumental knowledge
versus the formalism of abstraction? How do these views help us to address the dominance of automated thinking without simply proposing an inverted hierarchy or a flattening of theory and technique, ontology and history. That's my problem. And why, you may ask, these approaches you think are insufficient to address the computational stratum of media today? To reply to this question I need to quickly explain the deadlocks of these views, of what or I think at the dominant view, so at least this shifting view in what we call digital media theory today. On the one hand, Deleuze's proposition of an immanent thought
defies representational structure or cognitive identification in favor of the indeterminate being of the sensible. Mediation is not defined in terms of instrumental knowledge for Deleuze, but by the primacy of sensation. And that's, we're all familiar with affect theory. That's where it comes from. The transcendental excarnation from automated or habitual repetition that instead is as opposed to the unleashing of vibrations in thought. That's what the less really, that's pure thinking, pure vibration in thought. This platonic trace in the less is immanentism,
thus places philosophy, or what Deleuze and Gattari call conceptual persona, at the living threshold of theory and practice. Concepts are not symbolic ideas, for Deleuze, but unfold the energetic formation of ideas. This empirical transcendentalism precisely explains the virtual conditions of thinking in terms of causal indeterminacy found in biological or biophysical contingencies. So for the less is the reverberation of thought, the capacity for thought to think anew is pushed from below.
It's pushed from the vibration of matter. We are all familiar with the amount of work that has been produced in the last 10, 15 years within theories of affect or idea of vibration, vibrating matter. We know this stuff. We have inherited it already very clearly. On the other hand, Kittler espouses the Heideggerian apology for a no longer pure condition of being, mapping the historical decadence of theoretical knowledge, driven by the instrumental betrayal of sacred truth. The history of technology no longer allows theoretical thinking to have a privileged role.
Kittler indeed envisions that silicon-based binary language will soon be replaced by quantum computing, dissipating the limit between philosophy and automation for good. But these materialist approaches have distinct ontological implications. One proposes an imminent production of knowledge embedded in the abstraction of the sensible, while the other offers the historical ontology of technical knowledge. However, the conditions of mediation today are determined by a computational stratum whose capacity to organize and command information exceeds technical knowledge, but also the biophysical order of things.
This condition instead crucially involves the algorithmic automation of logic, whose learning skills are challenging the classical view of formal and universal language. This also means that algorithmic forecasting operates independently from biophysical organization, probability and chance. We cannot understand forecasting, computational forecasting, by extending biological understanding of noise, entropy, or probability of chance just coming from biology. We cannot do that because it's a different stratum. And that's for the sake of a sane universalism.
But I will tell you a bit later. To unpack computational forecasting, we then need to address the historical development, not of technique, but of the logic of technique, i.e. the computational abstraction of randomness. The materialist critique of media, as importantly rebrings, therefore, to the surface this tension between thinking and automation beyond biophysical substrates of being. Nevertheless, materialist approaches seem to overlook the scientific image of thought, and I borrow this idea from Wilford Sellers, American pragmatist, which importantly finds in the mechanization of logic
now satiated in machine learning an opportunity to individuate how technological abstraction has internally challenged the deductive logic of formal listening. So my point is that there is a challenge that computational automation by developing certain methods of reasoning has already posed to the classical understanding of formal reasoning in classic philosophy, i.e. based on formal reasoning. and theoretical reasoning only. But it will become clear, I hope, in a sec. Since the Turing machine, the condition of media can no longer exist
beyond computational abstraction, a mediatic condition overlooked by the critique of automated culture appealing to the non-conscious or irrational nature of instrumental knowledge. Instead, the consequences of the historical realization of logical thinking in automated systems has yet to be unpacked, and not many people want to go close to it. A materialist account of media needs to construct a non-representational, processual philosophy of computation that refuses any quick dismissal of logic, reason, or formal thought, or just think that as Heidegger, and this is also Kittler, inheritance of Heidegger,
that formal thought has come to an end. It is thus important with automation, and that's the critique that we hear constantly, the kind of critical theory of technology, that's my target. It is thus important to explain not simply the instrumental ground of algorithmic functions but the general functioning of an automated logic. So the general functioning of an automated logic. I, the patterning of non-inferential information from within computational processing in order to account for ontology of computation that is not just some kind of universal extension of what natural
law already do in physics and biology. In particular, the question of the limit of computation, being computable in the universal Turing machine, is for me the starting point for developing a critical theory of automation, which could account for the concretization of logical thinking and the abstraction of material information. A materialist approach to automation shall thus address the problem of the limit of computation in information theory to explain the use of randomness in the forecasting of probabilities that our automated system bring about today.
The incomputable has also been understood in terms of randomness and discussed in terms of the limits of reason. In the 1980s, information theorist Gregory Chaitin addressed the limit of computational logic from the standpoint of an entropic conception of information, intended as incompressible amounts of information. For Chaitin, computation corresponds to the algorithmic compression of maximally unknowable quantities of information, or maximal unknowable probabilities, an information process in which the disequilibrium between input and output shows the limits of formal logic.
Increasing quantity of information, also called algorithmic randomness, they market the point at which computation fails, the famous halting problem, as it stretches the logical procedure through which it was programmed and enters the randomness of informational substrates. Because of the tendency of information to increase in size, in volumes, and quantities within a system, algorithmic programming becomes the result of emerging patterns. This mechanization of logic in computational processing does expose the limits of formal logic. And precisely of logical deduction,
whose rules or truth universally banded all kinds of information. I can explain a bit more about deductive logic, but I hope it becomes clear in a sec. Instead for charting, computational processing breaks from the logic and importantly sets in place the incomputable condition for prediction increasing randomness in the system. Forecasting in computation depends not on given instruction leading to set probabilities. Instead, this form of statistical computation explains how incomputables are central to
the evolution of algorithmic patterns. This corresponds to an experimental method of prediction, whereby algorithmic order is not opposed to, but depends upon, infinite varieties of unknowns. Chaitin calls these computational dynamics experimental axiomatics. Here postulates, axiomatic truths, are result of the algorithmic compression of incomputables into patterns. Mediation here involves an algorithmic form of intelligibility, which grows patterns from within information environments,
undermining or confirming existing axiom rules and truths. So it's an experimental process of defining truth, as it were, as it were from the compression of information, rather than having rules and then compress information. In other words, with algorithmic information theory, we can suggest that if axioms are becoming experimental truth, That is, if laws are the result of an algorithmic intelligibility of information environments, then forecasting too may involve the process by which in computables become or come to be algorithmically printed, so printed from within automation itself.
But how to explain this non-deductive method without simply affirming lawless randomness disqualifies computational logic? The problem of what comes first and what becomes the principle, the ontological principle, is a question here. I think that the sublimation, so how to address this computational forecasting of unknowns without simply relinquishing logic in the name of contingency, randomness. I think that the sublimation, sublimation of indeterminacy, the sublimation of incomputable,
and randomness, which is also another level of intervention in digital theory, often reflected in critical theory, claims that automation perpetuates irrationality and stupidity, risks missing the fundamental epistemological shift of mechanization of logic. Aye, Jan. The latter, I suggest, importantly shows that the condition of mediation involves a new articulation of the relation between order and randomness in the iterative evolution of information which challenges some kind of safe distinction between theoretical
and practical reasoning. In other words, the rise of computational automata defines not the end of logic or the merging of logic and technique in instrumental knowledge but the realization that automation requires a continuous patterning of information the evolution or an algorithmic intelligibility of randomness ie the compression which is experimental experimenting exactly oh Oh, already? Oh, I have a lot to say. Sorry, I have to speed up. Experimenting with the conceptual infrastructure of truth.
More precisely, this involves an integrated function of labeling, selecting, evaluating, and commanding information, which increasing the skills of automated systems to incorporate social, economic, cultural practices within their vast parallel distributed level of iteration. So one can suggest that the automation of these practices amounts to a veritable computational stratification of discursive and non-discursive thinking. It is no longer possible to ignore the new order of intelligibility activated by these iterative compressions of randomness. The critique of techno-capitalist dominance of affective unconsciously rational thinking seems to overshadow the artificial evolution of logical reasoning
within computation and its experimental truth by simply merging logic and technique. Whilst acknowledging that the crisis of deductive logic and formal reasoning in computation has led to a critical reinvestment in instrumental knowledge, the logical reasoning of automatic needs to be addressed. Charles Sander Peirce, pragmatist conception of logic, offers us a possible perspective here. His efforts to define continuity between practical and theoretical knowledge resulted in the formulation of a triadic system of logic, what he called abductive, inductive, deductive inferential reasoning. His views did not reject but extended the
the Kantian model of deductive reasoning, and the transcendental function of the conceptual mediation for the prediction of truth. For Peirce, the logical structure of reasoning are embedded in material and non-inferential strata, but are not of the same order. Here, prediction is central to reasoning and involves this processual continuity between different orders of mediation, material, semiotic logical. In short, Peirce develops a materialist account of logic, the process by which hypothetical reasoning moves from non-inferential materialities to the articulation of discursive practice and to truths. Prediction is not simply a translation between objects
and concepts involving some kind of energetic transduction or code mutation. Instead, thinking itself involves prediction or hypothesis to the best explanation of things, exposing rather than imposing the conceptual infrastructure embedded within things. So the conceptual layers of things are, as it were, extrapolated by what he calls abduction and hypothetical thinking tracking truth retroactively. Prediction, therefore, is possible because hypothesis generates truth from physical and conceptual apprehension of the material layering of things. This involves the material layers of the logic of decision because it is where
information is extrapolated through a series of theoretical hypotheses that can be put to test and unfold the discursive practices and then construct systems of inferences. So in other words prediction is not without conception but concepts are as it were retroactivated from the material layers of information and evolution of inferential reasoning from non inferential practices so the the conceptual infrastructure of things from far from being directly accessible or innate within things rather needs to be made explicit it It needs to be articulated through a series of logical methods,
starting with hypotheses that are able to explain and transform truth according to material constraints. Axiom does become auto-generated rules because of the experimental processing that allows hypotheses to unfold the iterative loops of materialities where randomness becomes interior to truth. So randomness is not other than truth. is internal to truth. This dynamic logic admits that the attainment of truth is dependent on the function of reason, but reasoning implies the mediating activities of universal science, material, discursive, logical, where hypotheses serve to revise or confirm truth.
And I need to jump because I have two minutes. So, abduction, therefore, is not transduction. It privileges not the indeterminacy of causality, but the retroactive tracking of cause and its consequential truths. Transductive mediation admits an ontological condition of energetic influx, cutting across biological, physical, technical strata. a sort of monistic multiplicity in an invariant continuum of differences. Instead, abduction entails the process by which the limit of formal reasoning, or the incomputable in algorithmic information terms,
becomes the point at which non-inferential practices, i.e. material activities, practices, do not guarantee equivalence across scales, but are embedded in the elaboration of information into patterns and patterns into rules exposing some kind of multiscolored universe of order. In other words, abduction starts from collective non-inferential practices but then ends up articulating bounded patterns whose increasing skills in evolving new levels of order from randomness results in the establishment of rules. So, Peirce's triadic logic challenges both cognitive and affective model of thinking.
It offers a materialistic articulation of logic and explains the relation between thoughts and things in terms of a multi-scalor order of mediation, where randomness becomes internal to the iterative loop of information. The question is, what layering of randomness is occurring, at what level of computation? That's the research that has to be done. So my proposition is just an attempt to argue for the process by which the instrumental knowledge embedded in automated systems does not preclude that algorithmic patterns can exceed their practical functions. My point is that these functions do not simply accomplish tasks but grow hypotheses about
unknown layers of information and thus crucially mediate the technological abstraction of inferential reasoning. This will mean not the end of philosophy, as warned by Hitler. This means that the end of philosophy, as warned by Hitler, concerns not with the dystopic triumph of technical logic, of technical knowledge and the fallacy of theory, but with the effort to construct a general artificial reasoning. And that's the project of extended Kantian model of logic through Pierce, for which practical knowledge incorporates a conceptual infrastructure that algorithmic mediation is contributing to unpack and to evolve.
So mediation here involves a speculative forecasting of unknowns, insofar as the hypothetical causality within logic requires a continuity between practical and theoretical reasoning. Thank you.