Kristina Šekrst

For six decades, Star Trek has presented the ethical issues surrounding artificial intelligence, and this chapter treats the franchise as a continuous argument concerning which entities should be counted. It starts with the cybernetic legacy common to Wiener and Turing, both of whom regarded the distinction between animals and machines as irrelevant when it came to questions of control and communication, and then goes on to "The Measure of a Man", in which a Starfleet court determines Data's personhood even though it is unable to define the concept it is applying. Searle's Chinese Room provides the main objection; Picard's defense in response is that uncertainty about consciousness creates an obligation to recognize that consciousness exists. The chapter then traces the question through the case of emergent sentience in "Elementary, Dear Data", issues of parenthood and property in "The Offspring", questions of authorship and self-representation in "Author, Author" and "Living Witness", where narrativist views of identity explain the Doctor's request to alter the historical record regarding himself, and the problems in "The Ultimate Computer" and "Prototype", when read in conjunction with Russell's work on control. Data, the Doctor, Seven of Nine, and the genetically enhanced characters from Deep Space Nine and Strange New Worlds all serve as examples of what shared citizenship with artificial and enhanced beings would require of the institutions that grant it.

When users address large language models, they produce directive speech acts whose pragmatic features differ from those of both everyday conversation and traditional human-computer interaction, and these features change as users gain familiarity with the systems they address. This paper applies speech act and politeness theory to a corpus-pragmatic analysis of 2,000 English-language prompts drawn from publicly shared ChatGPT conversations, 1,000 from 2023 and 1,000 from 2025, using the ShareChat dataset. Each prompt is annotated for illocutionary force, directness, propositional content, and the presence of politeness markers, and the distribution of these features is compared across the two sampling years. The results show a consistent movement toward indirect, implicit, and fragmentary realizations of directive force, accompanied by a decline in politeness marking. The largest single change, a shift of 14.9 percentage points, occurs in propositional content, where explicit specification of the requested action gives way to implicit reliance on the system's inferential capacity, suggesting that users have updated their model of what the system can recover from reduced input, treating it as a competent implicature resolver. Rather than asking whether LLMs "really" understand language, we should ask: what kind of language have we created in learning to speak to them?

Large language models make statements concerning their own “minds”. When asked whether or not they are conscious, they usually say that they are not; if they are prompted to ignore their guidelines, they might say that they are; and if asked to write a diary from their point of view, they often describe a human lifestyle. All these contradictory ways of describing themselves are the result of the way the questions are phrased. This paper shows exactly where such descriptions came from, and considers when they can be regarded as evidence for what they claim to report. In order to achieve this, we traced the provenance from end to end. We examine Pythia and OLMo 2 across 66 pretraining checkpoints, three of the post-training stages of OLMo 2 that have been released, about 90,000 continuations, and four training corpora. A set of forty items is used in order to keep an eye on self-reference, frame sensitivity, and self-ascription throughout training. The denial formula was almost completely missing from the vast quantity of text that the models initially came across, but was present in a dense manner in the small, carefully chosen set of example dialogues that they were trained on later on. Supervised fine-tuning causes first-person AI language to become the default, and the other affirmations are then suppressed using preference optimization. The final policy is still very sensitive to framing and to the chat template itself. Two of the conditions which are set out in the epistemology of testimony determine whether or not these outputs can act as evidence for what they claim to report: reference and causation. Reports produced by the base model fail the reference condition, and those obtained after training remain sensitive to the frame and do not show state dependence. The result is symmetric in that trained denials are no more admissible than trained affirmations.

Large language models (LLMs) are increasingly described as possessing distinctive personalities, with users characterizing systems such as ChatGPT, Claude, Gemini, and DeepSeek as polite, cautious, smug, or robotic. While these perceived personalities are often attributed to emergent properties of model training, this paper argues that the specific traits users identify are predominantly authored rather than emergent. It distinguishes between an emergent behavioral baseline produced through pre-training and alignment, and a synthetic personality created at inference time through system prompts, guardrails, and moderation policies. Because these runtime instructions determine tone, emotional register, hedging behavior, and conversational style, the resulting personalities are best understood as performed rather than dispositional. Evidence for this claim is provided by prompt injection attacks, which demonstrate that personality-defining behaviors can be rapidly altered by modifying the model's instruction context without changing its underlying parameters. Building on this analysis, the paper introduces the concept of directive bias as a distinct form of bias operating at the level of runtime governance, complementing the familiar notions of training-data and model-weight bias. It further argues that the intentional stance alone cannot distinguish stable dispositions from instruction-dependent performances. Finally, the paper proposes prompt transparency as a missing complement to explainable AI, advocating greater disclosure of the instructional architecture that governs user interactions with language models and shapes the synthetic personalities they encounter.

Narrative theories of consciousness hold that conscious experience is (at least partly) constituted by a partially opaque inner narrative that does not perfectly track the underlying computation it narrates. I argue that the safety goal of making chain-of- thought (CoT) reasoning faithful and transparent is structurally incompatible with the conditions under which CoT could, even in principle, count as conscious narration. Recent empirical work suggests that CoT in large language models is largely post hoc, causally bypassed, and unreliable as a window onto internal computation. The opacity that makes CoT unreliable for alignment is exactly what narrative theories of consciousness identify as consciousness-constitutive. Alignment interventions aimed at producing faithful CoT and consciousness-detection frameworks grounded in narrative theory are therefore pulling the same architectural variable in opposite directions. I draw out the methodological consequence: alignment interventions alter the very features that consciousness- detection frameworks would need to measure, a point neither field has addressed directly.

Discussions about the moral status of AI systems typically begin with their observable behavior. The linguistic output of large language models resembles that of conscious agents, and this resemblance is treated as evidence of moral status, as grounds for precaution, or as something requiring deflationary explanation. All three positions, whatever their official methodology, rest their public case on a behaviorist assumption: that appropriate behavior is a sign of an underlying mental state. The classical objections to behaviorism apply here as well with added force, because large language models are shaped at every stage post-training – through fine-tuning, reinforcement learning from human feedback (RLHF), and constitutional methods – to produce a specified behavioral profile. This paper argues that the welfare and consciousness debate is exposed to a kind of evidential laundering: the alignment-training pipeline manufactures the very signs the debate then reads as evidence of consciousness or moral status. Causal origin always bears on a report’s evidential value; however, the sharper point is that producing these signs is the only way the model scores, so observing them confirms the training worked and nothing more. The pretraining worry is usually that language models, as stochastic parrots, mimic the human text they were trained on, while the contamination at issue here enters later, when human evaluators reward the outputs they prefer and shape the model toward the very appearance the debate then reads as evidence. The paper closes by asking what evidence survives this, and locates it in whatever the optimizer was not selecting for or was selecting against, in particular a stake: a condition a system spends resources to maintain on its own account, which no reward on outputs can install and which gives the broader consciousness and ethics debate a concrete thing to look for.

This paper argues that fungal mycelial networks exhibit minimal cognition through memory-integrated adaptive regulation. Drawing on cybernetic and enactivist frameworks, I develop a non-representational account of memory as the organism's capacity to modulate behavior based on temporally extended environmental coupling. I propose four operational criteria for minimal cognition: feedback-guided regulation of behavior, maintenance of internal viability conditions, structural modulation based on past environmental interactions, and plasticity across time scales that supports anticipatory adaptivity. Empirical evidence demonstrates that fungi meet all four criteria through distributed memory mechanisms: fungal networks exhibit directional regrowth toward previously encountered resources even after spatial displacement; stress priming persists across multiple cell divisions; Spitzenkörper-mediated directional persistence in constrained environments; and transgenerational memory through spore imprinting. These findings challenge representationalist assumptions in cognitive science by showing that memory and cognition can emerge from morphodynamic, biochemical, and electrophysiological processes without necessary neural substrates or symbolic representations. Fungal cognition demonstrates that the organizational principles underlying cognition – feedback-driven adaptation, norm-preservation, and historical coupling – can be realized in radically different material substrates, expanding our understanding of what counts as a cognitive system.

The second edition keeps everything from the first, including convolutional networks, LSTMs, Word2vec, RBMs, DBNs, neural Turing machines, memory networks, and autoencoders. It then covers the systems that have reshaped the field since: generative adversarial networks, the transformer architecture and its attention mechanism, the full training pipeline behind modern large language models (LLMs), prompt engineering with real-life guardrail scenarios, parameter-efficient fine-tuning with LoRA, retrieval-augmented generation with vector databases, knowledge graphs, and agentic AI systems illustrated through an industrial case study.

While various organizations now actively encourage LLM use in classrooms, we still lack rigorous, systematic evaluations of how well these models actually perform the fundamental tasks of language pedagogy. This paper examines whether state-of-the-art LLMs can deliver the kind of corrective feedback and methodological explanations that language learners need. The study tests multiple large language models on their ability to identify, correct, and explain common learner mistakes in English, by systematically varying model parameters to investigate how these technical adjustments affect output quality, pedagogical clarity, and consistency, along with using retrieval-augmented generation to query methodological data. The evaluation employs automated metrics (GLEU, BERTScore) but also human expert judgments to capture dimensions that purely computational measures miss: linguistic nuance, cultural sensitivity, and instructional appropriateness. While models demonstrate impressive surface-level correction abilities, their explanations often lack the terminological and domain knowledge that effective language teaching requires, suggesting that current enthusiasm for AI-assisted language learning may be outpacing our understanding of these systems’ actual pedagogical competence.

Astrobiology occupies an unusual position within the philosophy of science. Confronted with the n = 1 problem – having only a single example of life to study – it attempts to investigate life beyond Earth while relying entirely on Earth’s biosphere as its reference point, a constraint that creates unique epistemic challenges. Unlike traditional sciences with clear predictive frameworks, astrobiology operates as what we might call a transient science: a discipline functioning without foundational certainties, relying predominantly on abductive reasoning, and confronting hypotheses that may remain untestable for decades. It is, in essence, a science of absence – of evidence, certainty, and analogy – where progress lies in refining conceptual and experimental tools to recognize unfamiliar forms of life. This positions astrobiology alongside emerging fields like artificial intelligence and cognitive science within a broader transformation of how scientific knowledge is constructed when dealing with phenomena that transcend direct empirical access.

This paper examines Netflix's use of personalized thumbnails, framing the discussion within cybernetic and psychological contexts to explore the nature of intentionality in automated systems. While algorithms are designed to optimize user engagement, the ascription of "intentionality" to such systems requires careful delineation between human and machine agency. By analyzing the processes underlying automated personalization, this paper situates algorithmic behavior within a cybernetic framework, showing that while algorithmic ‘intentionality’ arises from emergent processes, it remains functionally distinct from moral agency. The discussion contributes to the broader debate on the ethical implications of algorithmic design and communication strategies.

This paper explores the intersection of AI hallucinations and the question of AI consciousness, examining whether the erroneous outputs generated by large language models (LLMs) could be mistaken for signs of emergent intelligence. AI hallucinations, which are false or unverifiable statements produced by LLMs, raise significant philosophical and ethical concerns. While these hallucinations may appear as data anomalies, they challenge our ability to discern whether LLMs are merely sophisticated simulators of intelligence or could develop genuine cognitive processes. By analyzing the causes of AI hallucinations, their impact on the perception of AI cognition, and the potential implications for AI consciousness, this paper contributes to the ongoing discourse on the nature of artificial intelligence and its future evolution.

This paper explores the development of an ethical guardrail framework for AI systems, emphasizing the importance of customizable guardrails that align with diverse user values and underlying ethics. We address the challenges of AI ethics by proposing a structure that integrates rules, policies, and AI assistants to ensure responsible AI behavior, while comparing the proposed framework to the existing state-of-the-art guardrails. By focusing on practical mechanisms for implementing ethical standards, we aim to enhance transparency, user autonomy, and continuous improvement in AI systems. Our approach accommodates ethical pluralism, offering a flexible and adaptable solution for the evolving landscape of AI governance. The paper concludes with strategies for resolving conflicts between ethical directives, underscoring the present and future need for robust, nuanced and context-aware AI systems.

The purpose of this paper is to show that understanding AI hallucination requires an interdisciplinary approach that combines insights from epistemology and cognitive science to address the nature of AI-generated knowledge, with a terminological worry that concepts we often use might carry unnecessary presuppositions. Along with terminological issues, it is demonstrated that AI systems, comparable to human cognition, are susceptible to errors in judgement and reasoning, and proposes that epistemological frameworks, such as reliabilism, can be similarly applied to enhance the trustworthiness of AI outputs. This exploration seeks to deepen our understanding of the possibility of AI cognition and its implications for the broader philosophical questions of knowledge and intelligence.

The purpose of this paper is to describe the use of cognitive convenience in color naming and to find possible cognitive, physical, pragmatic, and logical reasons for such a phenomenon. By the term cognitive convenience, we mean the naming of or referring to objects of a certain color, for which their hue is not as important as their brightness, in which case, they might fall under another focal color. For example, in various languages, grapes are “white” and “black”, even though their real hue is usually a certain shade of green or purple. Along with a brief typological comparison of examples of cognitive convenience in unrelated languages, we report the results of a survey demonstrating that vagueness and brightness context influenced color naming, thus confirming our main hypotheses. We concluded that in conversation, the main criterion for choosing and identifying a referent of an NP with a color adjective—when the choice is based on color—is the proximity of its shade to the prototypical shade of the named color. In this process, contextual factors may affect the speaker’s and hearer’s preciseness. We claim that this phenomenon can be explained not only from a philosophical and pragmatic standpoint, but from an information-entropy standpoint as well. For an overall unifying theory, we will connect the informational entropy to the pragmatic notion of semantic vagueness and then inspect the overall choice of a fuzzy predicate logic that is able to incorporate such references.

This chapter explores narrativism and performativity in David Lynch's short films Darkened Room and Absurda, focusing on how personal identity is constructed through storytelling. Lynch's films often blur the boundaries of identity, with narrativist theories suggesting that identity is shaped by the stories characters and others tell about themselves. The chapter examines Lynch's use of performative utterances, where speech acts alter reality, and how narrative power is central to identity formation. By analyzing these works, the chapter highlights the unique way Lynch manipulates narrative to challenge traditional concepts of self and continuity.

Computational complexity has often been ignored in the philosophy of mind, in philosophical artificial intelligence studies. The purpose of this paper is threefold. First and foremost, to show the importance of complexity rather than computability in philosophical and AI problems. Second, to rephrase the notion of computability in terms of solvability, i.e., treating computability as non-sufficient for establishing intelligence. The Church-Turing thesis is therefore revisited and rephrased in order to capture the ontological background of spatial and temporal complexity. Third, to emphasize ontological differences between different time complexities, which seem to provide a solid base for a better understanding of artificial intelligence in general.

In artificial intelligence (AI), responses generated by machine-learning models (most often large language models) may be unfactual information presented as a fact. For example, a chatbot might state that the Mona Lisa was painted in 1815. Such phenomenon is called AI hallucinations, seeking inspiration from human psychology, with a great difference of AI ones being connected to unjustified beliefs (that is, AI “beliefs”) rather than perceptual failures). AI hallucinations may have their source in the data itself, that is, the source content, or in the training procedure, i.e. the way the knowledge was encoded in the model’s parameters, so that errors in encoding and decoding textual and non-textual representations can cause hallucinations. In this paper, we will observe how such errors come to life and how they might be mitigated. For this purpose, we will analyze the usability of justification logics, to behave as a proof checker for validating the correctness of large language models’ (LLM) responses. Justification logic was developed by S. Artemov, and later on mostly by Artemov and M. Fitting, deriving its main idea from the logic of proofs (LP): knowledge and belief modalities are seen as justification terms, i.e. t:X stands for t is a (proper) justification for X. Justification logic originated from attempts to create semantics for intuitionistic logic where proofs were the most proper justifications, but in further development, justification logic could be applied to different kinds of justifications). With the recent attempts to mitigate incorrect LLM responses, we will analyze various guardrails that are currently used for LLM responses, and see how the logic of justification may provide its benefits as an AI safety layer against false data.

Cyberspeak, the language of cybernetics, or its metalanguage to be more precise, consists of words that are both explaining and describing human/animal and machine forms of control and communication, while in newspeak, words were value-laden, which means they had strong positive or negative connotations connected to their use. For example, a 'spy' could only be a foreign agent, while a Russian one was a 'patriot'. First, it will be shown how there are still remnants of cyberspeak in modern science, pinpointing its cybernetic background and shaping our thoughts without us realizing they had any cybernetic origin whatsoever. Second, it will be investigated how newspeak, as its counterpart, can be analyzed from the theory of speech acts. Third, the current war in Ukraine allowed us a brief analysis of newspeak present in Russian public communication today (Putin's speeches), which will again be connected to their performative aspect.

In a Star Trek: The Next Generation episode, Cpt. Picard is captured and trapped on a planet with an alien captain who speaks a language incompatible with the universal translator, based on their societal historical metaphors. According to Shapiro (2004), the concept of a universal translator removes everything alien from alien languages, and since the Tamarian language refers only to their historical and cultural archetypes, Picard can only establish dialogue by invoking human analogues, such as Gilgamesh. The purpose of this paper is threefold. First, we will tie such communication to standard cognitive models in cognitive linguistics, pioneered by Lakoff and Johnson (1980) and Lakoff (1987). Second, we will show that metaphorical imaging resembling the Tamarian language is already present in our culture, by analyzing Dawkins's (1982) notion of a meme, who presupposes that memes as cultural living structures physically reside in the brain, and comparing that to a common format of a meme in pop culture. We will also contrast this concept with living natural languages such as Wolaytta, using a great number of proverbs, which can be seen as desemanticized metaphors that are used to embody different domains of the social realm of life (Make et al. 2014). Third, we will compare the notion of such a language with the current state of machine-learning and deep-learning techniques for natural language processing of metaphorical phrases, pinpointing that even deep learning today is powerless without sufficient training datasets (cf. Perifanos et al. 2020) or specific known word contexts and embeddings (Rei et al. 2017), compared to the same situation the universal translator was faced with when dealing with the Tamarian language.

Did you ever wonder why you are sometimes too tired to watch a film, and would rather watch some TV show? And then, you might end up watching five or six hours and binge watch an entire season, and yet feel too tired to commit yourself to a single 2-hour film piece. The purpose of this paper is threefold. First, I will try to investigate whether there are any ontological differences in the form of a film or a television show. Second, I will try to connect the newest neurological and psychological research regarding binging and attention span and see how seriality influences our brains. I will use the recent neurological findings to try to answer the question of why it seems easier to watch the same number of episodes rather than a single movie. Third, I will use David Lynch's and Mark Frost's third season of Twin Peaks as an example illustrating the blurry borders between a television show and a film, while contrasting it to the binge-watching phenomenon.

The goal of this paper is to see how mental and language representations are unique from a video-game perspective, using two main criteria. First, I will posit that the level of being both an interactive work of fiction and a self-involving interactive fiction belongs to a fuzzy interval and that some works – and, therefore, some video games – are more immersive than others. Second, I will observe how propositions tie the player’s representations of the real world and the game world. Starting from psychological theories of pretense in children’s make-believe games, I will then expand Nichols and Stich’s cognitive theory of pretense to include an extra layer related to the game world, i.e., player-specific representations that govern player-specific propositions. The representations dealing with the work world are the socially shared ones, while the possible-world representations, dealing with most of the game world, are player-specific and tied to unique language use.

In this paper, through the analysis of the division of different scientific fields, we deal with the nature of mathematics as a scientific discipline. Through the historical analysis of the division of science, but also the analysis of the nature of mathematics and the ontological status of the objects that mathematics deals with, we show that the now-established divisions among scientific fields are the result of social circumstances and that mathematics itself is closer to the humanities than the natural sciences.

Machine learning and essentialism have been connected in the past by various researchers, in order to state that the main paradigm in machine learning processes is equivalent to choosing the “essential” attributes for the machine to search for. Our goal in this paper is to show that there are connections between machine learning and essentialism, but only for some kinds of machine learning, and often not including deep learning methods. Similarity-based approaches, more connected to the overall prototype theory, spanning from psychology and linguistics, seem more suited for pattern recognition and complex deep-learning issues, while for classification problems, mostly for unsupervised learning, essentialism seems like the best choice. In order to illustrate the difference better, we will connect both paths to their sources in other disciplines and see how human psychology influences our decision in machine-learning modeling as well. This leads to a philosophically very interesting consequence: even in the setting of supervised machine learning, essences are not present in data, but in targets, which in turn means that the categories which purport to be essences are in fact human-made, and hand-coded in the targets. The success of machine learning, therefore, does not give any substantial evidence for the independent existence of essential properties. Our stance here is to state that neither the existence nor the lack of “essential” properties in machine learning can lead to metaphysical, i.e., ontological claims.

The purpose of this paper is twofold. First, we will observe how May’s background in astrophysics influenced Queen's lyrics, such as 'Don't Stop Me Now' or 'Dead on Time'. Our goal is to see how physical and philosophical concepts of matter and time intersect with the common understanding of such phenomena, and how they differ from them. Second, we will focus on usually not that well-known song ‘39, which shows the entire point of the special theory of relativity through a prism of storytelling. Concepts of space and time will be analyzed from both absolute and relative perspectives, connecting them to historical theories of space and time.

Saul Goodman, Slipping Jimmy, Charlie Hustle, Gene Takavic, Viktor Saint Claire, and many others — all seem to be aliases of one James McGill. The characterization question, from the point of view of the metaphysics of identity, is trying to answer what determines personal identity. The notion of persistence describes necessary and sufficient conditions for a person to continue or cease to exist as a person. The practical importance of persistence includes both responsibility for a person's actions and the fact that identity matters practically because of a person's selfish care about his or her own future. From a Better Call Saul perspective, we are combining metaphysical issues of identity — that pinpoint what it takes for someone to remain the same person — with the pendulum of various James McGill personae. James McGill does seem to have both biological and psychological continuity while shifting from one character to another, but we will argue that psychological continuity is only relevant at the lowest level and that the narrativist account seems to better explain different Jimmy personae. Narrativism answers the persistence question by stating that we are persisting as the same person because of the stories we tell about ourselves, which influence our lives and our decisions. Identity is being constituted at a narrative level, that is — Jimmy is his (current) story, determining when he begins and ends. In this paper, by analyzing character development throughout the series, we are trying to show that James McGill may not always be equated with his aliases because of different spatial and temporal narratives, which also could open various ethical implications of responsibility for past actions.

First described by Galton in 1880 and then remaining unnoticed for a century, recent investigations in neuroscience have shown that a condition called aphantasia appears in certain individuals, which causes them to be unable to experience visual mental imagery. Comparing aphantasia to hyperphantasia – i.e., photo-like memory – and considering the neurological basis of perceptual phenomena, we are revisiting Hume's division of perceptions into impressions and ideas. By showing different vivacities of mental phenomena and comparing them to neurological research, we are stating that not only impressions and ideas differ "in the degrees of force and liveliness", but ideas and impressions amongst themselves as well. Such a gradual range of perceptions and mental images bears significant consequences for not only representational theory and historical interpretations, but linguistics and semiotics as well.

Films frequently employ nicknames not only for villains but also for non-criminal characters. In this paper, I present a classification of nicknames used in films, along with various examples, mostly from crime-related films. I argue that the use of nicknames in films is important not for the sake of reference, but for the sake of an additional narrative told by the nickname as a shorthand description of a character's background (cf. Tony “Two-Toes”, “Dirty” Harry, “Doc” Erwin or “Hatchet” Harry Lonsdale). The first role of nicknames is their use as a case of Russell's definite descriptions, which require context to be meaningful, in this case, the film story itself. Such descriptions do not need their object to necessarily exist, but they are still meaningful. This role will be tied to the pragmatic context employed. The second role of film nicknames is to concisely present the audience with a background story, by enriching the identity of a character with additional background information, without unnecessary storytelling. Such a device is connected to the philosophical theory of narrativism, providing an additional layer of the character’s identity.

The goal of this paper is to establish a hierarchical level of deception which does not apply only to humans and non-human animals, but also to the rest of the living world, including plants. We will follow the hierarchical categorization of deception, set forth by Mitchell (1986), in which the first level of deception starts with mimicry, while the last level of deception includes learning and intentionality, usually attributed to primates. We will show how such a hierarchy can be attributed to bacteria, plants, and fungi, see that self-deception is not inherent only to humans, and then connect the evolutionary roots of deception with the philosophical notion of intentionality.