Translating Franz Kafka’s “Josefine, the Singer or the Mouse People” as a Metaphor for AI

Abstract

Differing English translations of Franz Kafka’s “Josefine, the Singer or The Mouse People” have inspired diverse critical readings of the story. As a post-liminal text, a translation retrospectively highlights the ambiguity of the original’s rhetorical meaning. Read as a metaphor for artificial intelligence (A.I.), “Josefine” reflects an uncanny sort of regenerated reality as a conflicted narrator ponders the meaning of Josefine’s song following her disappearance. Likewise, the form produced by A.I. programs like ChatGPT following an initial human input is typically that of a narrative, albeit one devoid of creativity, replaced instead with algorithmic determinism. Philosophical questions about the discursive potential of technology such as generative A.I. pose challenges to the definitional assumptions about the form narrative takes in rhetorical situations, wherein the audience/reader is left with a message untethered from its prompter/writer.

1. The Problem of the Narrative Form as Presented by Franz Kafka

In the humanities, artificial intelligence (AI) presents a general problem to writers and in particular to those interested in rhetoric. In seconds, AI is capable of generating linguistic outputs that look and sound like an original human creation. Pulling from vast backlogs of data, generative AI responds to human inputs in ways that humans can easily understand, in any language, reproducing the familiar form of human discourse. However, a rhetorical problem is found in the unidirectionality of the AI output, because while its response to an initial human input is coherent, follow-up questions always remain unanticipated. Every human prompt to the AI is new, decontextualized from any sort of relationship that requires two sentient beings to gradually learn how to best communicate with one another in a given time and space. Since Aristotle, rhetoric has been a practice that takes place between more than one individual, a back-and-forth dialectic that results in persuading one another towards attainment of greater knowledge. Now, with AI, there is only one individual necessary for conversation, which is, of course, no conversation at all but still takes the form of discourse. What AI produces is an aimless linguistic form: stories that, in order to exist, require exigence without adherence to a rhetorical purpose or situation. To practice a rhetoric without purpose is to prime humans for results-based labor, rather than the process basis at the heart of dialectical interaction, wherein humans learn about each other’s expectations, boundaries, and congruencies through shared time and space. AI may serve as a useful tool for low-stakes text generation, but as much as it may look and sound like a human being, AI cannot participate in the rhetorical situation of the human being.

In “A Childhood Photograph”, Walter Benjamin writes, “It is possible to read Kafka’s animal stories for quite a while without realizing that they are not about human beings at all. When one encounters the name of the creature—monkey, dog, mole—one looks up in fright and realizes that one is already far away from the continent of man” (Benjamin 2019, p. 68). While Kafka’s “animal stories” may not be about human beings, there has never been any doubt that they were created by a human being. Contrastively, in the 21st century, stories may now be created by AI via generative technologies like ChatGPT, following human prompts. The form that these narratives take are grammatically perfect constructions, but technically they can only be sourced to a programmatic algorithm. Benjamin (2019) concludes, “But it is always Kafka; he divests the human gesture of its traditional supports and then has a subject for reflection without end” (Benjamin 2019, p. 68). Could something similar be said about the inverse situation, of the divesting that occurs when a non-human “author” writes in a distinctly human form? This paper will posit how the form of narrative, be it an output of a biological (as seen in the fictional work(s) of Kafka) or artificial intelligence (e.g., ChatGPT), contradicts the dialogic imperative of sociality, and, therefore, why proponents of the humanities should embrace rhetoric as a universal modality since it incurs a more dialectical sort of communication in a 21st century dominated by technocapitalism.

Franz Kafka’s last completed work, “Josefine, die Sängerin oder Das Volk der Mäuse”, was published in German in the Czech Prager Presse shortly after he moved back from Berlin to Prague, and less than two months before his death in 1924. The short story has an intertextual relation to Kafka’s other “animal stories”, which ambiguously anthropomorphize characters as having both human and non-human attributes. Two main voices are prominent in “Josefine”, one being the singer herself, from the third-person perspective, and the other a nameless narrator who provides selective details about Josefine, what she does, and how she fits into the society where she finds an audience (Gross 1985). I will refer to this narrator’s pronouns as they in this analysis for a few reasons. Firstly, they are a representative of a collective called the “mouse people”, and when the narrator speaks it is generally for the sake of this collective, not an individual self. Secondly, concerning how AI is concomitant with the narrator (more on this shortly), I do not personify AI as he or she but something closer to an aggregation of pronouncements. And finally, the use of plural pronouns with singular antecedents invites alternative possibilities for the reader to assign an identity to the narrator that is not tied to a binary definition. This is not to discredit the insights of Gross (1985) in Allen Thiher’s Franz Kafka: A Study of the Short Fiction (Thiher 1990) about the female Josefine’s contrast against a male antagonist but aims to suggest a reorientation of identity in the story that may emphasize the Other (Josefine) as alienated not merely by the norms of a patriarchy but rather by a post-human or non-human antagonism.

This first-person narrator’s perspective is constant throughout the story, but their reliability is questionable as they admonish Josefine one moment and praise her the next, as if they cannot authoritatively state how authentic Josefine’s “singing” truly is. Not unlike the paradoxically unknowing assuredness of AI, the narrator’s uncertainty forms the text’s antagonism toward its hero, who is self-assured in her duty to “sing” despite any opposition, a single-minded artist unable to devote herself to anything other than the most emphatic pfeifen. Kafka used this curious word in the original German to describe the sound that both Josefine and the “mouse people” make—the direct translation in English is “piping”.

When an English reader scans “Josefine”, they may be reading any number of potential translations from Kafka’s original German version, which mediately problematizes their potential interpretation of the story. Differing translations of the noise that Josefine makes when she performs have encouraged diverse critical readings of the story’s central allegory. The initial English version in the socialist Partisan Review (1942) creatively rendered Josefine’s pfeifen as “squeaking”, whereas Willa and Edwin Muir provided the directly literal translation of “piping” (Kafka et al. 1952) and Andrii Ponomarkenko used another variant, “whistling” (Kafka 2023), while Stanley Corngold’s 2007 translation returned to “squeaking”. Several recent scholars—including Naama Harel in her 2020 book (Harel 2020) about zoopoetics and Kata Gellen in her 2019 study of acousmatic sound—address how such distinctions affect the ambiguity of animality and sound in Kafka’s works, and how such vocal qualities specifically corroborate the theme of subject fragmentation in “Josefine”. The diversity of theories stretch into queer readings of Josefine as a femme version of Josef K. or even as a transvestite performer, such as in Wendy Zieler’s suggestion that the tale is “another … reading of the biblical Joseph, one that calls attention to the epicene/queer/flamboyant and performative aspects of Joseph’s character” (Zierler 2013). Going as far back as Max Brod’s reading, many interpret the “mice” in the story as allegorical representatives for the Jewish diaspora, as Zierler (2013) also summarizes: “Mark Anderson, in Kafka’s Clothes, and Sander Gilman, in Kafka: The Jewish Patient, additionally note the choice of mice for ‘Josephine’, given … the similarity between the German word maus and the verb mauscheln—to speak German ‘Mosaically’, like a Jew” (Zierler 2013, p. 104). Most broadly, though, scholars interpret the story as a self-referential metaphor for Kafka’s creative struggle as an artist. Thiher declares, “Kafka’s work itself is the kind of art that we feel at once we cannot live without, although his art demonstrates that we are not one step closer to any truth for having, like Josephine’s people, let this piping resound in our silence” (Thiher 1990). Herein lies a metaphor for AI, a contradictory silent but resounding echo of the language and the art of the human being. Thiher concludes that “this is a truth, too, and it is the truth of a paradoxical wisdom that makes of Kafka, if not of Josephine, one of the few truly great artists of the twentieth century” (Thiher 1990, p. 94). Can the same be said of AI-generated art, which only affirms its aesthetic output after the input of the human?

Kafka’s story about Josefine,1 with its multiple different translations, is a post-liminal text in the way it anticipates the relational incommensurability of a defined (written) sound that cannot be translated through an undefined subject, but is elevated by the necessarily complicating translation(s) into another language, which retrospectively highlights the inherent ambiguity of rhetorical meaning. This reading conceptualizes the variety and heterogeneity of the text and its relationship to multiple modes of understanding. A post-liminal text is one that points to a spatial and temporal passage through physical states of being, formally anticipating retrospection due to and because of a former experience of differentiation. Heterogeneity disrupts sameness in a post-liminal text; ambiguity and disjunction remain unresolvable and even necessary, inherently constraining a rhetorical knowledge of messages previously unknown to readers and writers, alike.

“Josefine, the Singer or The Mouse People” is a post-liminal text in two ways: extra-textually and textually. The composing of the work followed Kafka’s liminal experience as an impermanent resident of Berlin, where he, according to Diamant (2004) “wanted, more than ever, to live” since falling for his new love from the German capital (Diamant 2004). Leaving Prague for Berlin had long been a dream of Kafka’s: “when he was thirty, he had written in his diary, ‘If only it were possible to go to Berlin, to become independent, to live from one day to the next, even to go hungry, but to let all one’s strength pour forth’!” (Diamant 2004, p. 14), showing that the place itself was, for Kafka, symbolic of freedom from the bohemia of Czechia. But it was only after Berlin, once the liminal ontology informed his own personal metamorphoses as a human who not only could, but did, live in a different part of the world in a different type of way, that Kafka began writing “Josefine”. Kafka was aware that he was about to die of prolonged tuberculosis during the writing and publication of “Josefine”, as it was the imminent reason to return to his family in Prague. After receiving confirmation of his impending death and Kafka becoming intent to “defray the medical bills, he asked Brod to offer his last story … to their mutual friend Otto Pick, the literary editor at Prager Presse. ‘“Josephine” must help out a little, there’s no other way’, Kafka said”—this was another uniquely post-liminal moment in Kafka’s life, because the “request marked the first time Kafka urgently asked Brod to arrange for publication of his work, and Max leaped at the opportunity to help his friend—as did Otto Pick” (Diamant 2004, p. 94). This foresight informed the diametrical title of the text, as well as its themes, foreshadowing subjective instability in the structure and content of the narrative. Sattler (1977) is particularly adamant about the meaningfulness of this title amendment: “According to Max Brod’s account, one of Kafka’s last creative acts was to change the title of the story to include mention of the mouse people, thereby establishing clearly one aspect of the narrator’s point of view—he is obviously a mouse. This fact would be somewhat obscured, though by no means indiscernible, had the title only remained, ‘Josephine the Singer’” (Sattler 1977).

2. The Limits of Formalized Communication

Machine learning and generative AI can make no such anticipations about the potential successes or failures of the texts they produce, because they are nothing more than a “stochastic parrot”. Bender et al. (2021) point out how when the “risk connected to seeming coherence and fluency involves machine translation (MT) and the way that increased fluency of MT output changes the perceived adequacy of that output” (Bender et al. 2021) that this lack of intent for rhetorical use on the part of the AI (MT) will only repeat variations from its pre-programmed models, as a parrot squawking human-esque “language” merely mimics a meaningless sound. The parrot’s noise is nothing more than noise, even if in form it is recognizable to a human as a piece of human language. After the AI narrative has been generated, the conversation is over, or better yet, proves that it was never a conversation at all. Rhetorical situations require a message linking a living writer to a living reader, and vice versa, but ChatGPT is never a reader and always, perpetually, a non-living writer. AI does not read as a human reads, just as a parrot that squawks a human-sounding word does not speak the language of the human, just as the nameless narrator does not conceptualize the art of Josefine despite her inspiring song.

Having just passed through the liminal phase of a spatial sort (Berlin/the mouse people’s society) and looking ahead to yet another of a temporal sort (death/exile), the story of Josefine situates the impossible task of epistemological understanding from a position of difference/Other. Spivak et al. (1999) famously clarifies this postcolonial perspective: “Whether or not [the people] can see it … their text articulates the difficult task of rewriting its own conditions of impossibility as the conditions of its possibility” (Spivak et al. 1999). The journey of leaving home (“Umwelt”) and returning necessarily incurs the disruptive and discordant liminal zone, which can be experienced for a time but must not be held in a fixed state if a boundary or threshold is ever to be crossed. Josefine’s noise is remembered after she is gone, but scrutinized for its ambiguity. What kind of song could it be if it exists only in memory? If the creator of the art is no longer present, then what can those who remain possibly do with its untethered meaning?

The audience in a rhetorical situation is one which matters greatly, not only in compositional contexts, but also in translations (including machine translation), especially in adherence to the form of narrative, as Speirs et al. (2011) quote Phelan: “narrative can be fruitfully understood as a rhetorical act: somebody telling somebody else on some occasion and for some purpose(s) that something happened” (Speirs et al. 2011). (Emphasis my own to indicate necessity of retrospective interpretation.) Unless prompted, an AI translation will not adapt to a rhetorical situation, like the following variations on the title of Kafka’s text. The first indication of rhetorical disjunction from Kafka’s original, “Josefine, die Sängerin oder Das Volk der Mäuse”, leads to many English variations that suggest a multiplicity of different meanings. Firstly, in the Partisan Review as “Josephine, The Songstress or, The Mice Nation”, to the Muirs’, “Josephine, the Singer or the Mouse Folk”. Corngold changed even more: “Josefine, the Singer or The Mouse People”. The Partisan Review, a political publication that featured mainly socialist writings, translated “Volk” to “nation” which not only invokes a populist state power but also a proletariat. Known for publishing prominent anti-authoritarian voices from the mid-twentieth century such as George Orwell, this change appealed to the left-leaning, English-speaking audience of the publication. Other changes include an additional comma after “or”, and “Josephine” now with a specifically anglicized spelling and more recognizable biblical provenance (“Joseph”) as opposed to “Josef” [K, from The Trial?]. The “Songstress” is a direct translation from the gendered German form of “singer”, which indicates the female. A decade later, the Muirs made things more neutral with a less politically connotative “Folk”, a return to a single comma, and non-gendered “singer”. “Josephine” remained until Corngold returned to the original German name, “Josefine”. Not only does this originalist translation point to the intertextual cohesion with the protagonist from The Trial, but the pronunciation of the title is hybridized, as “Josefine” contains four syllables and a soft German accent, while “Josephine” sounds only three syllables and a hard “J” in its anglicized form. This title necessitates a double consciousness that recognizes two languages (two ways of understanding) at the same time. “People” was Corngold’s new choice for “Volk” as well, which begs the question of what Josefine and the narrator are. The only major term that is unchanging in every translation is “Mouse”, but, ironically, this is the most problematic noun in the title.

3. Categorizing Non-Human Ontologies for Human Purposes

The fixed binary meaning of a species category is disputable throughout the story—the “mouse” identity is never clearly metaphorical or literal, implying that a categorical (spatial) ambivalence resists certainty. Harel shows how Kafka’s particular German word choices are intentionally ambiguous whenever related to the qualitative type of creature that are the “mouse people”, using “Volk” instead of “Leute” or “Menschen”, for example (Harel 2020). Only once in the story does a variant of “mouse” appear, but it is a simile: “her audience does not squeak, we are quiet as mice” (Kafka 2007) [aber ihr Auditorium pfeift nicht, es ist mäuschenstill]. Unlike Brod’s interpretation, this is certainly not an “obvious” mouse in a literal sense, but the translation into English makes the characters more mouse-ish via connotations. The “fur” might be a fur coat, something worn to a musical performance in the 1920s, but if these are indeed mice with some anthropomorphic attributes, then the “fur” is simply these beings’ literal fur. The ambiguity of being thematically contrasts with becoming, correlated with the timing of how this also happens after Kafka has temporarily lived (liminally) in Berlin, and now prepares for death at home.

Comparatively, “Josefine” is Kafka’s sole “animal story” where it is unclear throughout whether the characters are human or not, making it distinct not only in the timeline of Kafka’s authorship as a last work, but also an only work. It is his only finished work written after his time in Berlin. In The Metamorphosis, even if it is unclear what sort of “vermin” [Ungeziefer] Gregor Samsa has transformed into, he is something like an “old dung beetle” [alter Mistkäfer] according to the charwoman (Kafka 2007, p. 71), but in “Josefine”, the beings are necessarily ambiguous (Harel 2020, p. 154). The confusion is the point. What are we, definitionally speaking? This is the deeper existential dilemma, complicated and compounded beyond the more basic representative question of who. Furthermore, how can we dare say who we are, until we have passed a threshold that offers a more suitable view from which to make an informed assessment? This is the artist’s position, post-liminally speaking. The experiencing of sense data as a neutral perception is not enough to make meaning, but requires subjective interpretation (Harel 2020, p. 155). Again, the latter is not possible for an AI program, because it is not a subject but a technological tool. It is only subjected, and always by a human. Following a negative definition of what it means to be human, we are not AI. Unlike AI, which writes unidirectionally, we are social, writing with purpose, for audiences, vacillating between reading and writing. Likewise, the narrator in “Josefine” does not comprehend the meaning of Josefine’s song after she is gone because her presence was all they could connect her message to as a metaphorical “mouse”, a rhetorical situation in which the audience is suddenly removed from the social element of communication, alienated from what before gave meaning to a community. Once the input is switched off and Josefine leaves her audience only a silent memory to work with, the discourse becomes one-sided, inert.

Diegetic sound in Kafka’s stories is something otherworldly if coming from a non-human creature. The narrator ponders a particularly ambiguous sound, which is central to their skepticism of Josefine’s merit: “Is it even song, then? Isn’t it perhaps just squeaking? And squeaking, of course, is something that all of us are familiar with” (Kafka 2007, p. 95) [Ist es denn überhaupt Gesang? Ist es nicht vielleicht doch nur ein Pfeifen? Und Pfeifen allerdings kennen wir alle]. Kafka chose the word “pfeifen” and not “piepsen”, which is the direct German translation for squeaking. Koelb (2019) offers some illuminating etymology here: “Even the term pfeifen that Kafka uses to designate Josephine’s art is drawn from the commonplace piping of gemeine Sprache…. Pfeifen is thus a contemptuous, abusive use of language, a kind of aggressive rhetoric that Kafka knew well… [A]ll of this piping is transformed by an animal pest (a mouse, a kavka) into an art that is arguably not art at all” (Koelb 2019). Squeaking is a mouse-like sound, but, more importantly, one that humans associate with mice. Mice do not call their sounds “squeaking”, but in their noise-making are producing something that is useful only for themselves, and humans do not understand this as mice do. Formalistically, a “pfeifen” cannot be heard but only read about, because it has no proper aurality for the human ear as something indicative of meaningful music. This sound is not a sound at all, but a description of an unmusical sound. Gellen says that this sound in the story is “theoretical” (Gellen Norberg 2019) and points to an anticipation of an experience not yet had in reality, e.g., the sound of death. AI has never been alive—the formal echo of human language that it produces is concomitant with the theoretical sound of the ambiguously written noise of Josefine. The perfect grammatical form it produces is meaningless, rhetorically.

An epistemology of ontology is post-liminal when more than one manner of living has been demonstrated, and the ending of “Josefine” shows how maintaining both ways simultaneously is not sustainable, following the law of liminality. The narrator is the voice throughout the story, even if Josefine has the voice. Or, she had the voice, since by the end things are spoken of less in the present tense and more as a recounting of what she did for the people in the past, up until she vanished, perhaps due to death: “the latest news, however, is that she had disappeared just at a moment when she was expected to sing” (Kafka 2018) [das Neueste aber ist, daß sie zu einer Zeit, wo ihr Gesang erwartet wurde, verschwunden war]. Likewise, two ways of reading oscillate in this text (in many layers), as mouse-people vs. people-mice, the narrator’s voice vs. Josefine’s voice, as present tense vs. past tense, Josefine’s genius vs. equivalence, the narrator’s admiration vs. antagonism, and so on. Driscoll (2017) situates Josefine’s “ambiguously singular-plural identity and her status as the singer of her community” (Driscoll 2017) as the story’s key to how “this voice does not found a new myth or a new homogeneous community, but is always at the limit (Nancy 1991, p. 67), and this limit can and must also be read as the ‘abyssal limit’ (Derrida 2008, p. 12) between ‘man’ and ‘animal’” (Driscoll 2017, p. 12). (Emphasis my own to implicate liminality within the formal structure of the story.) Or, in our 21st-century reading, “human” and “non-human” in the latter as a form of artificial intelligence. ChatGPT may appropriately be deemed the true “storytelling animal”. In Kafka’s story, ambivalence deviates from thematic homogeneity, as Harel says, “Josephine” “only offers constant vacillation between the two readings” (Harel 2020, p. 156). The vacillation creates the tension of double-consciousness—it comes with the knowledge of two ways of being, the dialectical process of becoming.

4. Rhetoric as a Discursive Antidote to Misunderstanding

Like a parrot mimicking human language, post-liminality incurs retrospection in the form of mimesis, or recalling the sameness of a thing after the repetition occurs as a reflection of a warped type of remembered reality, which is precisely what the narrator ponders about Josefine: “Was her actual squeaking notably louder and livelier than the memory of it will be?” (Kafka 2007, p. 107) [War ihr wirkliches Pfeifen nennenswert lauter und lebendiger, als die Erinnerung daran sein wird?]. As in a mirror, the human form may appear perfectly formed, but will, in its perfect semblance, contain no trace of human material, only glass. This is the problem of art: art is not only recurrence, nor is it a freezing of a frame, but a separation from present-tense reality that consciously comments on the impossibility of commenting on present-tense reality. Comay (2000) states, “According to a familiar Platonic formula… the uncontrolled reproducibility of the artwork expresses itself as an infinitely regressive mimetic flux. A genealogical catastrophe would have disordered the very process of reproduction. Copy and original become indistinguishable, the voice becomes a simulacrum of itself, the original no longer holds” (Comay 2000). The narrator is not the artist, but the artist is no longer present, so they wonder further, “Even during her lifetime was it ever more than a mere memory? Didn’t the people, in their wisdom, value Josefine’s song so highly precisely because in this way it could never be lost?” (Kafka 2007, p. 108) [War es denn noch bei ihren Lebzeiten mehr als eine bloße Erinnerung? Hat nicht vielmehr das Volk in seiner Weisheit Josefinens Gesang, eben deshalb, weil er in dieser Art unverlierbar war, so hoch gestellt?]. “Josefine” meta-textually comments on experiencing an artist’s art, from a voice that is bereft of artistic understanding, leaving only shards of some strange artifact. If the art remains with the audience, even after the artist has gone, the bestowing itself causes a fracture and lessening of the art.

In its perfect grammar, AI generates the ideal narrative form, which simultaneously undermines the human creators who filled up its language models with centuries of literature so that it might mimic it. Similarly, Vitzthum (1993) credits the nameless narrator as “the subverter of both the tradition and Josefine in that the very act of his writing undermines Josefine’s position as Singer” (Vitzthum 1993). Be it AI or the narrator, the passage across an unfamiliar threshold, a non-artist encountering art, changes the material of the aesthetic after the fact of artistic expression (a non-human generating human language is no different). However, due to the fragmentation, this material can only be commented on in ambivalence, and the ambivalence cannot be achieved without a passage beyond single-mindedness. AI works only in a single direction, discursively speaking, for it cannot read but will only ever “write”. Josefine is single-minded, but she cannot last as her story follows the law of narrative’s beginning, middle, and ultimate end. The narrator is conflicted because they remain, in the liminal space, left to reflect, “if it were true that Josefine does not sing but merely squeaks, and even perhaps, as it seems at least to me, barely exceeds the bounds of ordinary squeaking … then Josefine’s alleged artistry would be disproved” (Kafka 2007, p. 95) [Wenn es also wahr wäre, daß Josefine nicht singt, sondern nur pfeift und vielleicht gar, wie es mir wenigstens scheint, über die Grenzen des üblichen Pfeifens kaum hinauskommt … dann wäre zwar Josefinens angebliche Künstlerschaft widerlegt]. But Josefine does not merely squeak, because the sounds she makes are pfeifen.

One hundred years after Kafka’s death, humans have found themselves in the same situation with AI that the narrator was in with Josefine. The narrator’s output takes the form of a story, while Josefine’s form was that of song; she never was a storyteller. In their attempt to reproduce the meaningfulness of Josefine’s song in a different form, they confuse the meaning of her song in the story. Likewise, AI reproduces the form of narrative from past forms that were produced by humans, but AI can only offer responses in a single direction following the human input. Because AI is not human, the potential for confusion is always a risk with AI discourse. Whenever a human produces a narrative form, a key component is rhetorical purpose, which considers audience, but since AI can only blindly respond to human prompts and never via an inspired exigence to prompt a human (it never purposefully considers who its audience is), the result will be untrustworthy in any rhetorical situation.

The ethos (to recall Aristotle in conclusion) of AI will always be weak in discursive contexts. AI may answer a question that contains an objective answer, and may do so with correctness, but beyond this there is little value to prompting AI for creative tasks. What AI should mean to writers broadly and rhetoricians specifically, is what AI means operationally. The “artificial” part of AI is of prime importance to the human intelligence that programmed it, as AI makes perfect grammatical forms out of past forms, if and when prompted to do so. This sort of tool may be useful for neutral or unbiased “how to” instruction manuals for furniture, baking recipes, or driving directions, but until it has built furniture itself, baked a cake with a grandparent, or taken the scenic route with an old lover, the relational component that AI lacks will always prevent the technology from becoming a reliable source of inductive knowledge that can be meaningful. The narrator can never reproduce Josefine’s song in a narrative form, and any attempt to do so confuses the meaningfulness of her ephemeral art. AI can never reproduce human discourse, and any attempt to do so will confuse the purpose of a rhetorical situation. In a time when political and ideological divides are stark and growing ever wider, the need for human-to-human rhetoric is great, and no noise will communicate to a human better than that of the human being’s. Let us leave the AI to our own devices, and make it a device that is deductive and meaningless in its empty form, a resonance in Plato’s cave that echoes that we are still here, not for its purposes but for our own.