Philosophy of AI, Misc

This article cohesively discusses three topics, namely color and its perception, the yet-to-be-solved hard problem of consciousness, and the theoretical possibility of strong AI. First, the article restores color back into the physical world by giving cross-species evidence. Secondly, the article proposes a dual-field with function Q hypothesis (DFFQ) which might explain the ‘first-person point of view’ and so the hard problem of consciousness. Finally, the article discusses what DFFQ might bring to artificial intelligence and how it might allow strong (...) AI to stay true. (shrink)

The reasoning abilities of large language models (LLMs) are the topic of a growing body of research in artificial intelligence and cognitive science. In this paper, we probe the extent to which a dozen LLMs are able to distinguish logically correct inferences from logically fallacious ones. We focus on inference patterns involving conditionals (e.g., 'If Ann has a queen, then Bob has a jack') and epistemic modals (e.g., 'Ann might have an ace', 'Bob must have a king'). These inference patterns (...) have been of special interest to logicians, philosophers, and linguists, since they plausibly play a central role in human reasoning. Assessing LLMs on these inference patterns is thus highly relevant to the question of how much the reasoning abilities of LLMs match those of humans. Among the LLMs we tested, all but GPT-4 often make basic mistakes with conditionals. Moreover, even GPT-4 displays logically inconsistent judgments across inference patterns involving epistemic modals. (shrink)

The focus of this article is the "good-will theory", which explains the effect humans can have on the safety of AI, along with how it is in the best interest of a superintelligent AI to work alongside humans and not overpower them. Future papers dealing with the good-will theory will be published, but discuss different talking points in regards to possible or real objections to the theory.

Contains a description of a generalized and constructive formal model for the processes of subjective and creative thinking. According to the author, the algorithm presented in the article is capable of real and arbitrarily complex thinking and is potentially able to report on the presence of consciousness.

Chapter 1 of the ongoing online publication "Artificial Aesthetics: A Critical Guide to AI, Media and Design", Lev Manovich and Emanuele Arielli -/- Book information: Assume you're a designer, an architect, a photographer, a videographer, a curator, an art historian, a musician, a writer, an artist, or any other creative professional or student. Perhaps you're a digital content creator who works across multiple platforms. Alternatively, you could be an art historian, curator, or museum professional. -/- You may be wondering how (...) AI will affect your professional area in general and your work and career. Our book offers intellectual tools to help us better see the future of creative fields. These tools come from philosophy of art, experimental psychology, media theory, digital culture studies and data science. The book is the first to combine these different perspectives together. -/- We started the work on the book in summer 2019, exchanging numerous messages, commenting on each other ideas, and sharing drafts of sections. The final book is a result of this process. Although each chapter is written by one author, it reflects the discussions we had over 27 months. (shrink)

A simple argument suggests that we can fruitfully model advanced AI systems using expected utility theory. According to this argument, an agent will need to act as if maximising expected utility if they’re to avoid exploitation. Insofar as we should expect advanced AI to avoid exploitation, it follows that we should expected advanced AI to act as if maximising expected utility. I spell out this argument more carefully and demonstrate that it fails, but show that the manner of its failure (...) is instructive: in exploring the argument, we gain insight into how to model advanced AI systems. (shrink)

1. 1.1. Suffering is bad. This is why, ceteris paribus, there are strong moral reasons to prevent suffering. Moreover, typically, those moral reasons are stronger when the amount of suffering at st...

(Draft of Feb 2023, see upcoming issue for Chalmers' reply) In Reality+: Virtual Worlds and the Problems of Philosophy, David Chalmers argues, among other things, that: if we are living in a full-scale simulation, we would still enjoy broad swathes of knowledge about non-psychological entities, such as atoms and shrubs; and, our lives might still be deeply meaningful. Chalmers views these claims as at least weakly connected: The former claim helps forestall a concern that if objects in the simulation are (...) not genuine (and so not knowable), then life in the simulation is illusory and therefore, not as valuable as a non-simulated life. Taking up these questions, I argue that in general, the value of social knowledge for a meaningful life dramatically swamps the value of non-social knowledge for a meaningful life. Along the way, I propose a non-additive model of the meaningfulness of life, according to which the overall effect of some potential contributor of value to a life depends in part on what is already in a life. One upshot is that the vindication of non-social knowledge, absent a correlative vindication of social knowledge, contributes either not at all or scarcely at all to the claim that our lives in the simulation might be deeply meaningful. This is so even though the vindication of non-social knowledge does forestall the concern that in the simulation, our lives might be wholly meaningless. (shrink)

Linguistic outputs generated by modern machine-learning neural net AI systems seem to have the same contents—i.e., meaning, semantic value, etc.—as the corresponding human-generated utterances and texts. Building upon this essential premise, Herman Cappelen and Josh Dever's Making AI Intelligible sets for itself the task of addressing the question of how AI-generated outputs have the contents that they seem to have (henceforth, ‘the question of AI Content’). In pursuing this ambitious task, the book makes several high-level, framework observations about how a (...) meta-semantic account of the content of AI-generated outputs should proceed. (shrink)

Recent work on interpretability in machine learning and AI has focused on the building of simplified models that approximate the true criteria used to make decisions. These models are a useful pedagogical device for teaching trained professionals how to predict what decisions will be made by the complex system, and most importantly how the system might break. However, when considering any such model it’s important to remember Box’s maxim that "All models are wrong but some are useful." We focus on (...) the distinction between these models and explanations in philosophy and sociology. These models can be understood as a "do it yourself kit" for explanations, allowing a practitioner to directly answer "what if questions" or generate contrastive explanations without external assistance. Although a valuable ability, giving these models as explanations appears more difficult than necessary, and other forms of explanation may not have the same trade-offs. We contrast the different schools of thought on what makes an explanation, and suggest that machine learning might benefit from viewing the problem more broadly. (shrink)

In a recent paper, Erasmus et al. (2021) defend the idea that the ambiguity of the term “explanation” in explainable AI (XAI) can be solved by adopting any of four different extant accounts of explanation in the philosophy of science: the Deductive Nomological, Inductive Statistical, Causal Mechanical, and New Mechanist models. In this chapter, I show that the authors’ claim that these accounts can be applied to deep neural networks as they would to any natural phenomenon is mistaken. I also (...) provide a more general argument as to why the notion of explainability as it is currently used in the XAI literature bears little resemblance to the traditional concept of scientific explanation. It would be more fruitful to use the label “understandable AI” to avoid the confusion that surrounds the goal and purposes of XAI. In the second half of the chapter, I argue for a pragmatic conception of understanding that is better suited to play the central role attributed to explanation in XAI. Following De Regt (2017) and Kuorikoski and Ylikoski (2015), the conditions of satisfaction for understanding an ML system are fleshed out in terms of an agent’s success in using the system. (shrink)

Many ethical frameworks require artificial intelligence (AI) systems to be explainable. Explainable AI (XAI) models are frequently tested for their adequacy in user studies. Since different people may have different explanatory needs, it is important that participant samples in user studies are large enough to represent the target population to enable generalizations. However, it is unclear to what extent XAI researchers reflect on and justify their sample sizes or avoid broad generalizations across people. We analyzed XAI user studies (N = (...) 220) published between 2012 and 2022. Most studies did not offer rationales for their sample sizes. Moreover, most papers generalized their conclusions beyond their target population, and there was no evidence that broader conclusions in quantitative studies were correlated with larger samples. These methodological problems can impede evaluations of whether XAI systems implement the explainability called for in ethical frameworks. We outline principles for more inclusive XAI user studies. (shrink)

Forty-two years ago, Capra published “The Tao of Physics” (Capra, 1975). In this book (page 17) he writes: “The exploration of the atomic and subatomic world in the twentieth century has …. necessitated a radical revision of many of our basic concepts” and that, unlike ‘classical’ physics, the sub-atomic and quantum “modern physics” shows resonances with Eastern thoughts and “leads us to a view of the world which is very similar to the views held by mystics of all ages and (...) traditions.“ This article stresses an analogous situation in biology with respect to a new theoretical approach for studying living systems, Integral Biomathics (IB), which also exhibits some resonances with Eastern thought. Stepping on earlier research in cybernetics1 and theoretical biology,2 IB has been developed since 2011 by over 100 scientists from a number of disciplines who have been exploring a substantial set of theoretical frameworks. From that effort, the need for a robust core model utilizing advanced mathematics and computation adequate for understanding the behavior of organisms as dynamic wholes was identified. At this end, the authors of this article have proposed WLIMES (Ehresmann and Simeonov, 2012), a formal theory for modeling living systems integrating both the Memory Evolutive Systems (Ehresmann and Vanbremeersch, 2007) and the Wandering Logic Intelligence (Simeonov, 2002b). Its principles will be recalled here with respect to their resonances to Eastern thought. (shrink)

Prior work proposes a view of development of purpose and source of meaning in life as a more or less temporally distal project ideal self-situation in terms of which intermediate situations are experienced and prospects evaluated. This work considers Augustine on ensoulment alongside current work into self as adapted routines to common social regularities of the sort that Augustine found deficient. How can we account for such diversity of self-reported value orientation in terms of common structural dynamics differently developed, embodied (...) and enacted over the life course? This work in progress represents divergent views as different sorts of error theory, and hypothesizes that differential development of spindle neural projections underlies different accounts of error for attentive self-correction through iterative attenuation. Some implications for AI augmentation and value alignment of autonomous agents developed on such a model are briefly entertained. (shrink)

In his groundbreaking work, Abstraction and Empathy, Wilhelm Worringer delved into the intricacies of various abstract and figurative artworks, contending that they evoke distinct impulses in the human audience—specifically, the urges towards abstraction and empathy. This article asserts the presence of empirical evidence supporting the extension of Worringer’s concepts beyond the realm of art appreciation to the domain of art-making. Consequently, it posits that abstraction and empathy serve as foundational principles guiding the production of both abstract and figurative art. This (...) holds particular significance in the 21st century, where artificial intelligence (AI) assumes a creative role that was absent during Worringer’s initial formulation of his theory. Thus, this paper postulates that AI inherently harbors a predisposition for the generation of abstract art, owing to its non-living and inorganic origins and functioning. (shrink)

According to virtue epistemology, the main aim of education is the development of the cognitive character of students (Pritchard, 2014, 2016). Given the proliferation of technological tools such as ChatGPT and other LLMs for solving cognitive tasks, how should educational practices incorporate the use of such tools without undermining the cognitive character of students? Pritchard (2014, 2016) argues that it is possible to properly solve this ‘technology-education tension’ (TET) by combining the virtue epistemology framework with the theory of extended cognition (...) (EXT) (Clark and Chalmers, 1998). He argues that EXT enables us to consider tools as constitutive parts of the students’ cognitive system, thus preserving their cognitive character from technologically induced cognitive diminishment. The first aim of this paper is to show that this solution is not sufficient to solve the TET. Second, I aim to offer a complementary and more encompassing framework of tool-use to address the TET. Then, I apply it to the educational uses of ChatGPT as the most notable example of LLM, although my arguments can be extended to other generative AI systems. To do so, in Sect. 1.1, I present Pritchard’s framework of cognitive character and virtue epistemology applied in education, to which I am committed in this treatment. In Sects. 2 and 3, I respectively illustrate Pritchard’s (2014) solution to the TET, and I highlight the general limitations of his proposal. Thus, in Sect. 4.1 I characterize ChatGPT as a computational cognitive artifact using Fasoli’s (2017, 2018) taxonomy of cognitive artifacts. In Sect. 4.2, I introduce my proposal, which combines Pritchard’s account of virtue epistemology with Fasoli’s (2017, 2018) taxonomy of cognitive artifacts to address the TET. Finally, in Sect. 5.1, I present some epistemically virtuous uses of ChatGPT in educational contexts. To conclude, I argue in favor of a multidisciplinary approach for analyzing educational activities involving AI technologies such as ChatGPT. (shrink)

Many researchers and intellectuals warn about extreme risks from artificial intelligence. However, these warnings typically came without systematic arguments in support. This paper provides an argument that AI will lead to the permanent disempowerment of humanity, e.g. human extinction, by 2100. It rests on four substantive premises which it motivates and defends: first, the speed of advances in AI capability, as well as the capability level current systems have already reached, suggest that it is practically possible to build AI systems (...) capable of disempowering humanity by 2100. Second, due to incentives and coordination problems, if it is possible to build such AI, it will be built. Third, since it appears to be a hard technical problem to build AI which is aligned with the goals of its designers, and many actors might build powerful AI, misaligned powerful AI will be built. Fourth, because disempowering humanity is useful for a large range of misaligned goals, such AI will try to disempower humanity. If AI is capable of disempowering humanity and tries to disempower humanity by 2100, then humanity will be disempowered by 2100. This conclusion has immense moral and prudential significance. (shrink)

Advances in artificial intelligence (AI) create an increasing similarity between the performance of AI systems or AI-based robots and human communication. They raise the questions: whether it is possible to communicate with, understand, and even empathically perceive artificial agents; whether we should ascribe actual subjectivity and thus quasi-personal status to them beyond a certain level of simulation; what will be the impact of an increasing dissolution of the distinction between simulated and real encounters. (1) To answer these questions, the paper (...) argues that the precondition for actually understanding others consists in the implicit assumption of the subjectivity of our counterpart, which makes shared feelings and a „we-intentionality” possible. This assumption is ultimately based on the presupposition of a shared form of life, conceived here as „conviviality.” (2) The possibility that future artificial agents could meet these preconditions is refuted on the basis of embodied and enactive cognition, which links subjectivity and consciousness to the aliveness of an organism. (3) Even if subjectivity is in principle impossible for artificial agents, the distinction between simulated and real subjectivity might nevertheless become increasingly blurred. Here, possible consequences are discussed, especially using the example of virtual psychotherapy. Finally, the paper makes case for a mindful appproach to the language we use to talk about artificial systems and pleads for preventing a systematic pretense of subjectivity. (shrink)

Inspired by recent progress in multi-agent Reinforcement Learning (RL), in this work we examine the collective intelligent behaviour of theoretical universal agents by introducing a weighted mixture operation. Given a weighted set of agents, their weighted mixture is a new agent whose expected total reward in any environment is the corresponding weighted average of the original agents' expected total rewards in that environment. Thus, if RL agent intelligence is quantified in terms of performance across environments, the weighted mixture's intelligence is (...) the weighted average of the original agents' intelligences. This operation enables various interesting new theorems that shed light on the geometry of RL agent intelligence, namely: results about symmetries, convex agent-sets, and local extrema. We also show that any RL agent intelligence measure based on average performance across environments, subject to certain weak technical conditions, is identical (up to a constant factor) to performance within a single environment dependent on said intelligence measure. (shrink)

La scorciatoia - Come le macchine sono diventate intelligenti senza pensare in modo umano -/- Le nostre creature sono diverse da noi e talvolta più forti. Per poterci convivere dobbiamo imparare a conoscerle Vagliano curricula, concedono mutui, scelgono le notizie che leggiamo: le macchine intelligenti sono entrate nelle nostre vite, ma non sono come ce le aspettavamo. Fanno molte delle cose che volevamo, e anche qualcuna in più, ma non possiamo capirle o ragionare con loro, perché il loro comportamento è (...) in realtà guidato da relazioni statistiche ricavate da quantità sovrumane di dati. Eppure possono essere in certi casi più potenti di noi: ci osservano continuamente, e prendono decisioni al nostro posto. E allora come incorporarle nella nostra società senza rischi ed effetti collaterali? Questo libro - rigoroso, pungente, originale nell'approccio - ci spiega come siamo arrivati sin qui, e indica il percorso che ci aspetta prima di poterci fidare di questi nuovi agenti «alieni». La tecnologia non basta, occorre un dialogo tra scienze naturali e umane: è il passaggio cruciale per una convivenza sicura con questa nuova forma di intelligenza. (Il Mulino). (shrink)

There has been a push in recent years to provide better explanations for how AIs make their decisions. Most of this push has come from the ethical concerns that go hand in hand with AIs making decisions that affect humans. Outside of the strictly ethical concerns that have prompted the study of explainable AIs (XAIs), there has been research interest in the mere possibility of creating XAIs in various domains. In general, the more accurate we make our models the harder (...) they are to explain. Go playing AIs like AlphaGo and KataGo provide fantastic examples of this phenomenon. In this paper, I discuss a non-exhaustive list of the leading theories of explanation and what each of these theories would say about the explainability of AI-played moves of Go. Finally, I consider the possibility of ever explaining AI-played Go moves in a way that meets the four principles of XAI. I conclude, somewhat pessimistically, that Go is not as imminently explainable as other domains. As such, the probability of having an XAI for Go that meets the four principles is low. (shrink)

Mental representations remain the central posits of psychology after many decades of scrutiny. However, there is no consensus about the representational format(s) of biological cognition. This paper provides a survey of evidence from computational cognitive psychology, perceptual psychology, developmental psychology, comparative psychology, and social psychology, and concludes that one type of format that routinely crops up is the language-of-thought (LoT). We outline six core properties of LoTs: (i) discrete constituents; (ii) role-filler independence; (iii) predicate–argument structure; (iv) logical operators; (v) inferential (...) promiscuity; and (vi) abstract content. These properties cluster together throughout cognitive science. Bayesian computational modeling, compositional features of object perception, complex infant and animal reasoning, and automatic, intuitive cognition in adults all implicate LoT-like structures. Instead of regarding LoT as a relic of the previous century, researchers in cognitive science and philosophy-of-mind must take seriously the explanatory breadth of LoT-based architectures. We grant that the mind may harbor many formats and architectures, including iconic and associative structures as well as deep-neural-network-like architectures. However, as computational/representational approaches to the mind continue to advance, classical compositional symbolic structures – that is, LoTs – only prove more flexible and well-supported over time. (shrink)

Can we build machines with which we can have interesting conversations? Observing the new optimism of AI regarding deep learning and new language models, we set ourselves an ambitious goal: We want to find out how far we can get in creating a digital replica of a philosopher. This project has two aims; one more technical, investigating of how the best model can be built. The other one, more philosophical, explores the limits and risks which are accompanied by the creation (...) of digital replicas. In cooperation with Daniel Dennett, we took his complete works to fine-tune a GPT-3. In this paper, we share our first results with some piloting models and discuss legal and ethical questions about creating and interacting with digital replicas. (shrink)

We present a game mechanic called pseudo-visibility for games inhabited by non-player characters (NPCs) driven by reinforcement learning (RL). NPCs are incentivized to pretend they cannot see pseudo-visible players: the training environment simulates an NPC to determine how the NPC would act if the pseudo-visible player were invisible, and penalizes the NPC for acting differently. NPCs are thereby trained to selectively ignore pseudo-visible players, except when they judge that the reaction penalty is an acceptable tradeoff (e.g., a guard might accept (...) the penalty in order to protect a treasure because losing the treasure would hurt even more). We describe an RL agent transformation which allows RL agents that would not otherwise do so to perform some limited self-reflection to learn the training environments in question. (shrink)

Since the beginning of the 21st century, technologies like neural networks, deep learning and “artificial intelligence” (AI) have gradually entered the artistic realm. We witness the development of systems that aim to assess, evaluate and appreciate artifacts according to artistic and aesthetic criteria or by observing people’s preferences. In addition to that, AI is now used to generate new synthetic artifacts. When a machine paints a Rembrandt, composes a Bach sonata, or completes a Beethoven symphony, we say that this is (...) neither original nor real art, but simply the complex imitation and reproduction of existing products of human culture. We face the old question concerning the nature of creativity: what kind of recombination of ideas, unusual analogies, and conceptual connections are considered the mark of originality? Can AI produce artworks? Could machines reach a point at which we consider them genuinely creative? We also need to investigate the challenges posed by AI-art to the notion of authoriality: Who is the author of an artificially generated artifact? An artificial system could be considered just an artist’s and a programmer’s tool. However, we are also fascinated by the idea of autonomous artificial creativity in the aesthetic domain, as a manifestation of highly intelligent behavior. This paper will try to define some key questions around what could it mean to consider a machine creative or even equipped with artistic intentionality. (shrink)

This paper further cashes out the notion that particular types of intelligent systems are susceptible to the is-ought problem, which espouses the thesis that no evaluative conclusions may be inferred from factual premises alone. Specifically, it focuses on top-down artificial moral agents, providing ancillary support to the view that these kinds of artifacts are not capable of producing genuine moral judgements. Such is the case given that machines built via the classical programming approach are always composed of two parts, namely: (...) a world model and utility function. In principle, any attempt to bridge the gap between these two would fail, since their reconciliation necessitates for the derivation of evaluative claims from factual premises. (shrink)

Addressing biases in natural-language processing (NLP) systems presents an underappreciated ethical dilemma, which we think underlies recent debates about bias in NLP models. In brief, even if we could eliminate bias from language models or their outputs, we would thereby often withhold descriptively or ethically useful information, despite avoiding perpetuating or amplifying bias. Yet if we do not debias, we can perpetuate or amplify bias, even if we retain relevant descriptively or ethically useful information. Understanding this dilemma provides for a (...) useful way of rethinking the ethics of algorithmic bias in NLP. (shrink)

It is relatively easy to state that information retrieval is a scientific discipline but it is rather difficult to understand why it is science because what is science is still under debate in the philosophy of science. To be able to convince others that IR is science, our ability to explain why is crucial. To explain why IR is a scientific discipline, we use a theory and a model of scientific study, which were proposed recently. The explanation involves mapping the (...) knowledge structure of IR to that of well-known scientific disciplines like physics. In addition, the explanation involves identifying the common aim, principles and assumptions in IR and in well-known scientific disciplines like physics, so that they constrain the scientific investigation in IR in a similar way as in physics. Therefore, there are strong similarities in terms of the knowledge structure and the constraints of the scientific investigations between IR and scientific disciplines like physics. Based on such similarities, IR is considered a scientific discipline. (shrink)

In a recent article, Erasmus, Brunet, and Fisher (2021) argue that Artificial Neural Networks (ANNs) are explainable. They survey four influential accounts of explanation: the Deductive-Nomological model, the Inductive-Statistical model, the Causal-Mechanical model, and the New-Mechanist model. They argue that, on each of these accounts, the features that make something an explanation is invariant with regard to the complexity of the explanans and the explanandum. Therefore, they conclude, the complexity of ANNs (and other Machine Learning models) does not make them (...) less explainable. In this reply, it is argued that Erasmus et al. left out one influential account of explanation from their discussion: the Unificationist model. It is argued that, on the Unificationist model, the features that makes something an explanation is sensitive to complexity. Therefore, on the Unificationist model, ANNs (and other Machine Learning models) are not explainable. It is emphasized that Erasmus et al.’s general strategy is correct. The literature on explainable Artificial Intelligence can benefit by drawing from philosophical accounts of explanation. However, philosophical accounts of explanation do not settle the problem of whether ANNs are explainable because they do not unanimously declare that explanation is invariant with regard to complexity. (shrink)

Feminist philosophers have long emphasized the ways in which women’s oppression takes a variety of forms depending on complex combinations of factors. These include women’s objectification, dehumanization and unjust gendered divisions of labour caused in part by sexist ideologies regarding women’s social role. This paper argues that feminized artificial intelligence (feminized AI) poses new and important challenges to these perennial feminist philosophical issues. Despite the recent surge in theoretical and empirical attention paid to the ethics of AI in general, a (...) strikingly small portion of this work has considered the impact of feminized AI in particular. Our aim is thus to draw attention to the new and pressing ethical challenges posed by feminized AI. Doing so is crucial not just to understanding the impact of these increasingly ubiquitous technologies, but also to our understanding of longstanding feminist philosophical concerns and efforts to ameliorate them. (shrink)

As more objections have been raised against grant peer-review for being costly and time-consuming, the legitimate question arises whether machine learning algorithms could help assess the epistemic efficiency of the proposed projects. As a case study, we investigated whether project efficiency in high energy physics can be algorithmically predicted based on the data from the proposal. To analyze the potential of algorithmic prediction in HEP, we conducted a study on data about the structure and outcomes of HEP experiments with the (...) goal of predicting their efficiency. In the first step, we assessed the project efficiency using Data Envelopment Analysis of 67 experiments conducted in the HEP laboratory Fermilab. In the second step, we employed predictive algorithms to detect which team structures maximize the epistemic performance of an expert group. For this purpose, we used the efficiency scores obtained by DEA and applied predictive algorithms – lasso and ridge linear regression, neural network, and gradient boosted trees – on them. The results of the predictive analyses show moderately high accuracy, indicating that they can be beneficial as one of the steps in grant review. Still, their applicability in practice should be approached with caution. Some of the limitations of the algorithmic approach are the unreliability of citation patterns, unobservable variables that influence scientific success, and the potential predictability of the model. (shrink)

Theologians often reflect on the ethical uses and impacts of artificial intelligence, but when it comes to artificial intelligence techniques themselves, some have questioned whether much exists to discuss in the first place. If the significance of computational operations is attributed rather than intrinsic, what are we to say about them? Ancient thinkers—namely Augustine of Hippo (lived 354–430)—break the impasse, enabling us to draw forth the moral and metaphysical significance of current developments like the “deep neural networks” that are responsible (...) for some of the most remarkable achievements of the past few years. First, Augustine’s theology of the natural world’s rationes seminales makes sense of neural networks’ success by explaining the world as a kaleidoscopic refraction of divine Wisdom rather than as a collection of discrete objects standing in crisp relations. Second, his account of interpretive judgment as a moral act bound up with love reveals how our training and use of AI relates to the Christian’s assimilation to Wisdom. The contingent meaning of the neural network reveals AI’s potential either for standing between us and the world or in some sense facilitating a Christian’s regathering of the created echoes of divine Wisdom throughout his or her journey into the Trinity. (shrink)

Recent research into the nature of self in artificial and biological systems raises interest in a uniquely determining immutable sense of self, a “metaphysical ‘I’” associated with inviolable personal values and moral convictions that remain constant in the face of environmental change, distinguished from an object “me” that changes with its environment. Complementary research portrays processes associated with self as multimodal routines selectively enacted on the basis of contextual cues informing predictive self or world models, with the notion of the (...) constant, per-vasive and invariant sense of self associated with a multistable attractor set aiming to ensure personal integrity against threat of disintegrative change. This paper proposes that an immutable sense of self emerges as a global attractor which can be described as a project ideal self-situation embodied in frontal medial processes during more or less normal adolescent development, and that thereafter serves to orient agency in the more or less free development of embodied potentials over the life course in effort to realize project conditions, phenomenally identified with the felt pull towards this end as purpose of and source of meaning in life. So oriented, life-long self-development aims to embody solutions to problems at different timescales depending on this embodied purpose, ultimately in the service of evolutionary processes securing organism populations against threats of disintegrative change over timespans far beyond that of the individual. After characterizing the target sense of self, research circling this target is briefly surveyed. Self as global project and developmental neural correlates are proposed. Then, the paper discusses some implications for research in biological and artificial systems. Building from earlier work in cognitive neurorobotics, discussion affirms the value of reinforcement rituals including prayer in metaphysical self-development, considers implications for value alignment and rights associated with free will in the context of artificial intelligence and robot religion, and concludes by emphasizing the importance of self-development toward project ideals as source of meaning in life in the current social-political environment. (shrink)

In 2011, Hibbard suggested an intelligence measure for agents who compete in an adversarial sequence prediction game. We argue that Hibbard’s idea should actually be considered as two separate ideas: first, that the intelligence of such agents can be measured based on the growth rates of the runtimes of the competitors that they defeat; and second, one specific (somewhat arbitrary) method for measuring said growth rates. Whereas Hibbard’s intelligence measure is based on the latter growth-rate-measuring method, we survey other methods (...) for measuring function growth rates, and exhibit the resulting Hibbard-like intelligence measures and taxonomies. Of particular interest, we obtain intelligence taxonomies based on Big-O and Big-Theta notation systems, which taxonomies are novel in that they challenge conventional notions of what an intelligence measure should look like. We discuss how intelligence measurement of sequence predictors can indirectly serve as intelligence measurement for agents with Artificial General Intelligence (AGIs). (shrink)

The workshop “Machine Learning: Prediction Without Explanation?” brought together philosophers of science and scholars from various fields who study and employ Machine Learning (ML) techniques, in order to discuss the changing face of science in the light of ML's constantly growing use. One major focus of the workshop was on the impact of ML on the concept and value of scientific explanation. One may speculate whether ML’s increased use in science exemplifies a paradigmatic turn towards mere pattern recognition and prediction (...) and away from science’s traditional aim of explanation. In contrast, certain conceptions of explanation, such as statistical explanation, could turn out to fit well with the achievements of present-day ML and concede an explanatory value to these achievements after all. It is an open question how to explain ML successes themselves, and this question was in the focus of several talks in the workshop. Based on the topics raised, we will discuss the talks’ contents in more detail as organized into (i) practitioners’ perspectives, (ii) explanations from ML, (iii) explanations of ML, (iv) societal implications, and (v) global and historical perspectives. (shrink)

We propose that, for the purpose of studying theoretical properties of the knowledge of an agent with Artificial General Intelligence (that is, the knowledge of an AGI), a pragmatic way to define such an agent’s knowledge (restricted to the language of Epistemic Arithmetic, or EA) is as follows. We declare an AGI to know an EA-statement φ if and only if that AGI would include φ in the resulting enumeration if that AGI were commanded: “Enumerate all the EA-sentences which you (...) know.” This definition is non-circular because an AGI, being capable of practical English communication, is capable of understanding the everyday English word “know” independently of how any philosopher formally defines knowledge; we elaborate further on the non-circularity of this circular-looking definition. This elegantly solves the problem that different AGIs may have different internal knowledge definitions and yet we want to study knowledge of AGIs in general, without having to study different AGIs separately just because they have separate internal knowledge definitions. Finally, we suggest how this definition of AGI knowledge can be used as a bridge which could allow the AGI research community to import certain abstract results about mechanical knowing agents from mathematical logic. (shrink)

We define a notion of the intelligence level of an idealized mechanical knowing agent. This is motivated by efforts within artificial intelligence research to define real-number intelligence levels of compli- cated intelligent systems. Our agents are more idealized, which allows us to define a much simpler measure of intelligence level for them. In short, we define the intelligence level of a mechanical knowing agent to be the supremum of the computable ordinals that have codes the agent knows to be codes (...) of computable ordinals. We prove that if one agent knows certain things about another agent, then the former necessarily has a higher intelligence level than the latter. This allows our intelligence no- tion to serve as a stepping stone to obtain results which, by themselves, are not stated in terms of our intelligence notion (results of potential in- terest even to readers totally skeptical that our notion correctly captures intelligence). As an application, we argue that these results comprise evidence against the possibility of intelligence explosion (that is, the no- tion that sufficiently intelligent machines will eventually be capable of designing even more intelligent machines, which can then design even more intelligent machines, and so on). (shrink)

We study the structure of families of theories in the language of arithmetic extended to allow these families to refer to one another and to themselves. If a theory contains schemata expressing its own truth and expressing a specific Turing index for itself, and contains some other mild axioms, then that theory is untrue. We exhibit some families of true self-referential theories that barely avoid this forbidden pattern.

After generalizing the Archimedean property of real numbers in such a way as to make it adaptable to non-numeric structures, we demonstrate that the real numbers cannot be used to accurately measure non-Archimedean structures. We argue that, since an agent with Artificial General Intelligence (AGI) should have no problem engaging in tasks that inherently involve non-Archimedean rewards, and since traditional reinforcement learning rewards are real numbers, therefore traditional reinforcement learning probably will not lead to AGI. We indicate two possible ways (...) traditional reinforcement learning could be altered to remove this roadblock. (shrink)

En era de la inteligencia artificial (IA) no han sido pocos los que se han preguntado si una máquina puede crear arte. En este sentido, la investigadora cognitiva Margaret Boden (2011) ha definido un tipo especial de arte al relacionar los conceptos "creatividad" y "computación". Así, el arte generado por computador (computer-generated art) es “the artwork results from some computer program being left to run by itself, with minimal or zero interference from a human being” (p. 141). Uno de los (...) casos más populares de este tipo de arte es el cuadro Portrait of Edmond de Belamy (2018), ampliamente publicitado como la primera obra de arte de una IA en ser subastada en Christie’s (Still & d´Inverno, 2019). Que las máquinas creen arte no ha dejado indiferente al mundo artístico. En efecto, se ha tendido a criticar al CG-art porque carecería de la inspiración mística que tiene el arte humano. Como menciona Aaron Hertzmann (2018), científico principal de Adobe Research San Francisco, “it is as if only humans create art because only humans have ‘souls.’ Surely, there should be a more scientific explanation” (p. 1). Atendiendo a lo anterior, sostengo que el CG-art tiene un valor estético descifrable a través de dos aproximaciones, a saber, evaluación humana y maquinal (Arriagada, 2020). (shrink)

This book provides a critical reflection on automated science and addresses the question whether the computational tools we developed in last decades are changing the way we humans do science. More concretely: Can machines replace scientists in crucial aspects of scientific practice? The contributors to this book rethink and refine some of the main concepts by which science is understood, drawing a fascinating picture of the developments we expect over the next decades of human-machine co-evolution. The volume covers examples from (...) various fields and areas, such as molecular biology, climate modeling, clinical medicine, and artificial intelligence. The explosion of technological tools and drivers for scientific research calls for a renewed understanding of the human character of science. This book aims precisely to contribute to such a renewed understanding of science. (shrink)

Reasoning about analogical arguments is known to be subject to a variety of cognitive biases, and a lack of clarity about which factors can be considered strengths or weaknesses of an analogical argument. This can make it difficult both to design empirical experiments to study how people reason about analogical arguments, and to develop scalable tutoring tools for teaching how to reason and analyze analogical arguments. To address these concerns, we describe WG-A (Warrant Game — Analogy), a framework for people (...) to analyze analogical arguments based on Bartha’s (2010) Articulation Model of analogical argumentation. We carry out two experiments designed to probe WG-A’s effectiveness in improving participants’ ability to reason about analogical arguments and argumentation in general, and argue that WG-A is a promising approach, though it is in need of further development. (shrink)

As the demand for explainable deep learning grows in the evaluation of language technologies, the value of a principled grounding for those explanations grows as well. Here we study the state-of-the-art in explanation for neural models for natural-language processing (NLP) tasks from the viewpoint of philosophy of science. We focus on recent evaluation work that finds brittleness in explanations obtained through attention mechanisms.We harness philosophical accounts of explanation to suggest broader conclusions from these studies. From this analysis, we assert the (...) impossibility of causal explanations from attention layers over text data. We then introduce NLP researchers to contemporary philosophy of science theories that allow robust yet non-causal reasoning in explanation, giving computer scientists a vocabulary for future research. (shrink)

Legg and Hutter, as well as subsequent authors, considered intelligent agents through the lens of interaction with reward-giving environments, attempting to assign numeric intelligence measures to such agents, with the guiding principle that a more intelligent agent should gain higher rewards from environments in some aggregate sense. In this paper, we consider a related question: rather than measure numeric intelligence of one Legg- Hutter agent, how can we compare the relative intelligence of two Legg-Hutter agents? We propose an elegant answer (...) based on the following insight: we can view Legg-Hutter agents as candidates in an election, whose voters are environments, letting each environment vote (via its rewards) which agent (if either) is more intelligent. This leads to an abstract family of comparators simple enough that we can prove some structural theorems about them. It is an open question whether these structural theorems apply to more practical intelligence measures. (shrink)

In their thought-provoking paper, Legg and Hutter consider a certain abstrac- tion of an intelligent agent, and define a universal intelligence measure, which assigns every such agent a numerical intelligence rating. We will briefly summarize Legg and Hutter’s paper, and then give a tongue-in-cheek argument that if one’s goal is to become more intelligent by cultivating music appreciation, then it is bet- ter to use classical music (such as Bach, Mozart, and Beethoven) than to use more recent pop music. The (...) same argument could be adapted to other media: books beat films, card games beat first-person shooters, parables beat dissertations, etc. We leave it to the reader to decide whether this argument tells us something about classical music, something about Legg-Hutter intelligence, or something about both. (shrink)

New developments in the field of communication and information technology will profoundly reshape the answers to questions of deep interest for humanity and philosophy. Who are we and what kind of relationship we establish among us? The boundaries between real life and virtual life tend to evanish. We are progressively becoming part of a global “infosphere”. This candid interview with professor Floridi try to shed some light on these issues, by considering the philosophical framework developed by the “infosphere philosopher”.

Purpose of the article is to identify the religious factor in the teaching of transhumanism, to determine its role in the ideology of this flow of thought and to identify the possible limits of technology interference in human nature. Theoretical basis. The methodological basis of the article is the idea of transhumanism. Originality. In the foreseeable future, robots will be able to pass the Turing test, become “electronic personalities” and gain political rights, although the question of the possibility of machine (...) consciousness and self-awareness remains open. In the face of robots, people create their assistants, evolutionary competition with which they will almost certainly lose with the initial data. For successful competition with robots, people will have to change, ceasing to be people in the classical sense. Changing the nature of man will require the emergence of a new – posthuman – anthropology. Conclusions. Against the background of scientific discoveries, technical breakthroughs and everyday improvements of the last decades, an anthropological revolution has taken shape, which made it possible to set the task of creating inhumanly intelligent creatures, as well as changing human nature, up to discussing options for artificial immortality. The history of man ends and the history of the posthuman begins. We can no longer turn off this path, however, in our power to preserve our human qualities in the posthuman future. The theme of the soul again reminded of itself, but from a different perspective – as the theme of consciousness and self-awareness. It became again relevant in connection with the development of computer and cloud technologies, artificial intelligence technologies, etc. If a machine ever becomes a "man", then can a man become a "machine"? However, even if such a hypothetical probability would turn into reality, we cannot talk about any form of individual immortality or about the continuation of existence in a different physical form. A digital copy of the soul will still remain a copy, and I see no fundamental possibility of isolating a substrate-independent mind from the human body. Immortality itself is necessary not so much for stopping someone’s fears or encouraging someone’s hopes, but for the final solution of a religious issue. However, the gods hold the keys to heaven hard and are unlikely to admit our modified descendants there. (shrink)

In this paper I argue that the search for explainable models and interpretable decisions in AI must be reformulated in terms of the broader project of offering a pragmatic and naturalistic account of understanding in AI. Intuitively, the purpose of providing an explanation of a model or a decision is to make it understandable to its stakeholders. But without a previous grasp of what it means to say that an agent understands a model or a decision, the explanatory strategies will (...) lack a well-defined goal. Aside from providing a clearer objective for XAI, focusing on understanding also allows us to relax the factivity condition on explanation, which is impossible to fulfill in many machine learning models, and to focus instead on the pragmatic conditions that determine the best fit between a model and the methods and devices deployed to understand it. After an examination of the different types of understanding discussed in the philosophical and psychological literature, I conclude that interpretative or approximation models not only provide the best way to achieve the objectual understanding of a machine learning model, but are also a necessary condition to achieve post hoc interpretability. This conclusion is partly based on the shortcomings of the purely functionalist approach to post hoc interpretability that seems to be predominant in most recent literature. (shrink)

As we increasingly delegate decision-making to algorithms, whether directly or indirectly, important questions emerge in circumstances where those decisions have direct consequences for individual rights and personal opportunities, as well as for the collective good. A key problem for policymakers is that the social implications of these new methods can only be grasped if there is an adequate comprehension of their general technical underpinnings. The discussion here focuses primarily on the case of enforcement decisions in the criminal justice system, but (...) draws on similar situations emerging from other algorithms utilised in controlling access to opportunities, to explain how machine learning works and, as a result, how decisions are made by modern intelligent algorithms or 'classifiers'. It examines the key aspects of the performance of classifiers, including how classifiers learn, the fact that they operate on the basis of correlation rather than causation, and that the term 'bias' in machine learning has a different meaning to common usage. An example of a real world 'classifier', the Harm Assessment Risk Tool (HART), is examined, through identification of its technical features: the classification method, the training data and the test data, the features and the labels, validation and performance measures. Four normative benchmarks are then considered by reference to HART: (a) prediction accuracy (b) fairness and equality before the law (c) transparency and accountability (d) informational privacy and freedom of expression, in order to demonstrate how its technical features have important normative dimensions that bear directly on the extent to which the system can be regarded as a viable and legitimate support for, or even alternative to, existing human decision-makers. (shrink)