Chinese Room Research Articles

Strong and weak alignment of large language models with human values

Minimizing negative impacts of Artificial Intelligent (AI) systems on human societies without human supervision requires them to be able to align with human values. However, most current work only addresses this issue from a technical point of view, e.g., improving current methods relying on reinforcement learning from human feedback, neglecting what it means and is required for alignment to occur. Here, we propose to distinguish strong and weak value alignment. Strong alignment requires cognitive abilities (either human-like or different from humans) such as understanding and reasoning about agents’ intentions and their ability to causally produce desired effects. We argue that this is required for AI systems like large language models (LLMs) to be able to recognize situations presenting a risk that human values may be flouted. To illustrate this distinction, we present a series of prompts showing ChatGPT’s, Gemini’s and Copilot’s failures to recognize some of these situations. We moreover analyze word embeddings to show that the nearest neighbors of some human values in LLMs differ from humans’ semantic representations. We then propose a new thought experiment that we call “the Chinese room with a word transition dictionary”, in extension of John Searle’s famous proposal. We finally mention current promising research directions towards a weak alignment, which could produce statistically satisfying answers in a number of common situations, however so far without ensuring any truth value.

The Realism of Speculative Fiction: Planetary Polyphony and Scale in Kim Stanley Robinson’s The Ministry for the Future

In “Thinking in SF Mode” (2021), Isabelle Stengers distinguishes between philosophical thought experiments (such as Maxwell’s demon [1867] or Searle’s Chinese room) and the ability of speculative fiction to test hypothesizes in fully developed storyworlds, arguing for the heuristic power of fictional density and immersion. We push this hypothesis further by examining what embodied reading brings to the political and ecological exploration of terraforming (geoengineering) in Kim Stanley Robinson’s Mars Trilogy (1992–1996).

ChatGPT language recognition limitations Prompts in Hungarian in a network of Latin and Cyrillic letters (Netlinguistics case study. The “Chinese room” experiment)

ChatGPT language recognition limitations Prompts in Hungarian in a network of Latin and Cyrillic letters (Netlinguistics case study. The “Chinese room” experiment)

A filosofia da mente como um problema da filosofia e da ciência. Trans/form/ação: revista de filosofia da Unesp, Marília, v. 47, n. 2, “Perspectivas femininas no pensamento filosófico”, e0240025, 2024.

Abstract: The purpose of the scientific article is a comparative analysis of the philosophy of consciousness in the context of Western and Eastern theory. In the process of the three stages of scientific research, the authors used general scientific and unique methods, such as analysis, synthesis, generalization, induction, comparative, and content analysis methods. The Western and Eastern philosophy representatives’ scientific and methodological works are analyzed to explain the problem of human consciousness. The features of Islamic monotheism in Al-Farabi’s views, as well as the relationship between thinking and body in Ibn Sina’s theory, are considered. The interpretation of the philosophy of consciousness of the Western region is presented on the example of Dennett’s heterophenomenology and physicalism, the “Chinese room” thought experiment by Searle and others. The fundamental similarities and differences in explaining the philosophy of consciousness by Western and Eastern scientists are determined. The author’s analytical criticism of these philosophical trends is provided.

The Possibility of Artificial Qualia

The discussions about qualia are mind-body problems. If artificial intelligence provides a functionally identical body, will it possess the same conscious experience? The possibility of artificial qualia, which is on the cutting edge of the development of artificial intelligence, will be the main topic of this essay. This paper argues the presupposition of the possibility of artificial qualia is the possibility of strong AI. Searl was the first person to question the possibility of strong AI by formulating the Chinese Room Argument. This paper introduces common objections from artificial intelligence practitioners and the development of modern AI, proving the invalid of the objection and claiming the possibility of strong AI. Next, this paper follows David Chalmers argument about the principle of organizational invariance and reconstructs Chalmers two Reductio ad Absurdum arguments to rebut objections. By formulating fading qualia argument and dancing qualia argument, the absent qualia argument and the inverted qualia argument are proved wrong. The principle can support the possibility of artificial qualia.

In Stanisław Lem's Orbit

In Stanisław Lem's Orbit

Can the Chinese Room Speak? Language and Racialized Posthuman Agency in Ex Machina

Can the Chinese Room Speak? Language and Racialized Posthuman Agency in Ex Machina

On Chatbots and Generative Artificial Intelligence.

On Chatbots and Generative Artificial Intelligence.

Grounding the Vector Space of an Octopus: Word Meaning from Raw Text

Most, if not all, philosophers agree that computers cannot learn what words refers to from raw text alone. While many attacked Searle’s Chinese Room thought experiment, no one seemed to question this most basic assumption. For how can computers learn something that is not in the data? Emily Bender and Alexander Koller (2020) recently presented a related thought experiment—the so-called Octopus thought experiment, which replaces the rule-based interlocutor of Searle’s thought experiment with a neural language model. The Octopus thought experiment was awarded a best paper prize and was widely debated in the AI community. Again, however, even its fiercest opponents accepted the premise that what a word refers to cannot be induced in the absence of direct supervision. I will argue that what a word refers to is probably learnable from raw text alone. Here’s why: higher-order concept co-occurrence statistics are stable across languages and across modalities, because language use (universally) reflects the world we live in (which is relatively stable). Such statistics are sufficient to establish what words refer to. My conjecture is supported by a literature survey, a thought experiment, and an actual experiment.

Знание-что, знание-как, сознание и искусственный интеллект

The article is devoted to the discussion about the relationship between knowledge and skill. This discussion arose within the critique of dualism. It is believed that dualism presupposes an intellectualism in which knowledge-how always derived from knowledge-that. The article proposes a method of distinguishing knowledgehow and knowledge-that through characteristic properties. This method allows us to draw a strict boundary between the two types of knowledge. This is followed by an explanation of the relationship between the two types of knowledge through the inclusion of knowledge-how as a necessary but not sufficient condition for knowledge-that. In the final part of the article, the principle of indistinguishability of knowledge-how among different agents is introduced, on the basis of which it is argued that the division into knowledge-how and knowledge-that works in the interests of dualism, and not vice versa, as supposed by Gilbert Ryle. This division makes sense in light of the Chinese room argument and the demonstration of the difference between humans and weak artificial intelligence.

ÇAĞDAŞ ZİHİN FELSEFESİNDE YAPAY ZEKÂ TARTIŞMALARINA BİR GİRİŞ: TAKLİTÇİLİK OYUNU VE ÇİNCE ODASI ARGÜMANI

Bu çalışma, yapay zekâ alanı teknik bir disiplinken söz konusu disiplinin felsefede nasıl gündeme geldiğini açıklamayı amaçlamaktadır. Çalışmanın temel iddiası, Alan Turing’in taklitçilik oyunu ile beraber yapay zekâ konusunun dil becerisi üzerinden insanın zihinsel kapasitelerini taklit etmeyi amaçlayan bir disipline dönüşmüş olduğudur. Yapay zekâ, akıllı çıktılar veren makineler veya sistemler üretmekle ilgilenen bir bilgisayar bilimi dalıdır. Bu dal; mantık, matematik, mühendislik alanlarıyla etkileşimli olarak doğmuştur. Başlangıcından sonra kat ettiği aşamalar ile farklı disiplinlerle ilişkilenmiştir. Söz konusu disiplinlerden birisi de felsefedir. Bu çalışma, yapay zekâ konusuna çağdaş zihin felsefesi cihetinden bir giriş yapmayı hedeflemektedir. Çalışmada yapay zekâ konusunun felsefe alanında gündeme gelmesi, “Makineler düşünebilir mi?” sorusunu tartışan Alan Turing’in taklitçilik oyunu üzerinden sunulmuştur. Bu bağlamda yapay zekâ tartışmalarının felsefi anlamda başlangıcı, dil becerisi üzerinden makineler ve insanların kıyaslanmasına dayandırabilir. Söz konusu teste birtakım eleştiriler sunulmuştur. Bunlardan biri John. R. Searle’ün Çince odası argümanıdır. Çalışmada bahsi geçen argümanlar sunularak yapay zekâ konusunun felsefi açıdan değerlendirilmesi amaçlanmıştır.

The Difficulties in Symbol Grounding Problem and the Direction for Solving It

The symbol grounding problem (SGP) proposed by Stevan Harnad in 1990, originates from Searle’s “Chinese Room Argument” and refers to the problem of how a pure symbolic system acquires its meaning. While many solutions to this problem have been proposed, all of them have encountered inconsistencies to different extents. A recent approach for resolving the problem is to divide the SGP into hard and easy problems echoing the distinction between hard and easy problems for resolving the enigma of consciousness. This however turns out not to be an ideal strategy: Everything related to consciousness that cannot be well-explained by present theories can be categorized as a hard problem which as a consequence would doom the SGP to irresolvability. We therefore argue that the SGP can be regarded as a general problem of how an AI system can have intentionality, and develop a theoretical direction for its solution.

The paradox of denial and mystification of machine intelligence in the Chinese room

Two critical questions spun the web of the Turing test debate. First, can an appropriately programmed machine pass the Turing test? Second, is passing the test by such a machine, ipso facto, considered proof that it is intelligent and hence “minded”? While the first question is technological, the second is purely philosophical. Focusing on the second question, this article interrogates the implication of John Searle’s Chinese room denial of machine intelligence. The thrust of Searle’s argument is that a machine lacks intentionality, so it can only simulate intelligence, not duplicate it. In his thinking, whatever a machine inputs to generate an intelligent output has no bearing on humanlike intelligence. Incidentally, Searle did not classify such a machine’s output as simulated and non-intelligent, nor did he explain how this output is actualised with mere simulation. The connection between “unintelligent machine’s input” and “intelligent machine’s output” is at this point shrouded in mystery. Consequently, the more Searle attempts a denial of machine intelligence in the Chinese room, the more he mystifies it. Using the method of critical analysis, this article advances three fundamental arguments to prove a machines’ obscurity in the Chinese room thought experiment. On the ground of Searle’s conviction, the first argument queries the absurdity in bypassing intentionality to produce intelligence; the second points out the obfuscation in generating intelligence with mere computation, and the third draws attention to the dilemma of classifying a machine’s output either as real-life intelligent behaviour or simulated intelligent behaviour.

Psychometric evaluation of the perceived perioperative competence scale-revised among the Chinese operating room nurses: a methodological research

BackgroundPerioperative competence is necessary to evaluate operating room nurses. The Perceived Perioperative Competence Scale-Revised (PPCS-R) is the only available tool developed specifically for the perioperative setting. However, there is a lack of research on the reliability and validity of this scale among Chinese nurses. Thus, the aim of this study is to translate, culturally adapt, and evaluate the psychometric properties of the Perioperative competence Scale-Revised (PPCS-R) among operating room nurses in China.MethodsInstrument cultural adaptation was carried out through forward translation, back translation, expert panel evaluation and pretesting. The psychometric properties (content validation, item analysis, construct validation, and reliability coefficient) of the Chinese PPCS-R (C-PPCS-R) were examined. An online survey was completed from June to August 2020 by operating room nurses (N = 480) in five third-grade class-A hospitals in Beijing.ResultsThe item analysis identified six items for scale reduction. Exploratory factor analysis showed the remaining 34 items loaded on six factors, which were named in accordance with the original scale. The six-factor model showed a good fit through confirmatory factor analysis. The item content validity index for the C-PPCS-R items ranged from 0.857 to 1.000, and that for the total scale was 0.875. Cronbach’s alpha was showed 0.787. Together, the six factors explained 68.62% of the variance.ConclusionsThe 34-item C-PPCS-R showed good validity and reliability to measure perceived competence among operating room nurses in the Chinese context. The scale can assist nurse managers to identify operating room nurses’ perceived competence, and provides evaluation criteria for career planning, performance appraisal, job assignment, and continuing education.

Data Analysis in an Entanglement Network Using Numerical Methods

While the foundation of quantum theory has been debated, its pragmatism has made it enormously productive. Establishing secret keys over long distances has been realized in the real world. Once considered only a hype, quantum computers have also been implemented in laboratories and are performing computations that are superior to their classical counterparts. In this paper, building on previous work, three 2-level entangled states are studied. In particular, the extensive range of states that yield the near-optimal result when entanglement swapping is applied at joints is numerically examined. This result is useful in establishing long-distance maximally entangled states, which are often preferred to short, non-maximal ones when used in applications. The precise nature of physical reality has been debated ever since the birth of quantum theory about a century ago. In this paper, reality is described not only in its physical aspects, but also as it pertains to consciousness. This physical reality in the context of mind is discussed using various examples, including entanglement and the Chinese room argument.

IS TURING TEST STILL PROFICIENT AND OPERATIVE AT PRESENT STATE OF THE ART?: BEYOND TURING TEST FOR THE NEXT GENERATION AI FRAMEWORKS

The problem of consciousness in terms of artificial intelligence is a difficult and big problem. With the test he put forward, the efficiency of artificial intelligence was discussed and tested. Some scientists have criticized the inability to distinguish between humans and robots with the Turing Test. Problems such as how sufficient this is and how it is possible to compare the intelligence of a human with the intelligence of a robot have been handled philosophically. The main purpose of this article is to address the adequacy of Turing testing and to question artificial intelligence tests and tools that can shed light on shaping the design of next-generation AI architectures. Searle's Chinese room experiment has been reconsidered by Turing by addressing the subjectivity-objectivity problem of Qualia philosophers and giving place to criticisms that can be directed to this test and countercriticisms that can be made to these criticisms. In addition, the role of the new generation Turing test in modeling concepts such as artificial consciousness and machine self-awareness and evaluating their performance is discussed.

The Gramophone: Lem against the Chinese Room

The text focuses on Lem’s rejection of the Chinese Room, a prominent challenge to the sufficiency of the Turing test. After outlining Lem’s relationship to the Turing test, it offers an exposition of two of Lem’s thought experiments, the Gramophone and the Jigsaw, whose critique is directly related to the critique of the Chinese Room. The text shows that Lem’s key argument is to point out the computational naivety of the machines that feature in these experiments. The text concludes by presenting some of Lem’s views on the nature of machine consciousness.

Dismantling the Chinese Room with linguistic tools: a framework for elucidating concept-application disputes

AbstractImagine advanced computers that could, by virtue merely of being programmed in the right ways, act, react, communicate, and otherwise behave like humans. Might such computers be capable of understanding, thinking, believing, and the like? The framework developed in this paper for tackling challenging questions of concept application (in any realm of discourse) answers in the affirmative, contrary to Searle’s famous ‘Chinese Room’ thought experiment, which purports to prove that ascribing such mental processes to computers like these would be necessarily incorrect. The paper begins by arguing that the core issue concerns language, specifically the discourse-community-guided mapping of phenomena onto linguistic categories. It then offers a model of how people adapt language to deal with novel states of affairs and thereby lend generality to their words, employing processes of assimilation, lexemic creation, and accommodation (in intersense and intrasense varieties). Attributions of understanding to some computers lie in the middle range on a spectrum of acceptability and are thus reasonable. Possible objections deriving from Searle’s writings require supplementing the model with distinctions between present and future acceptability, and between contemplated and uncontemplated word uses, as well as a literal-figurative distinction that is more sensitive than Searle’s to actual linguistic practice and the multiplicity of subsenses possible within a single literal sense. The paper then critiques two misleading rhetorical features of Searle’s Chinese Room presentation, and addresses a contemporary defense of Searle that seems to confront the sociolinguistic issue, but fails to allow for intrasense accommodation. It concludes with a brief consideration of the proper course for productive future discussion.

Düşünce Deneyleri Doğallaştırıldı

Thought experiments have long been utilized in science and philosophy. They come in various forms, from Schrödinger's cat in physics to the Chinese room argument in philosophy. Although thought experiments are used widely, scant attention has been paid to how thought experiments actually work. In philosophy literature, it has been argued that thought experiments are a disguised form of meaning analysis. Thought experiments rely on the notion of analyticity on their basis, and therefore they should be abandoned. In this paper, I argue that if we distinguish conceivability type thought experiments commonly found in philosophy from factive thought experiments widely used in science, a viable methodological alternative can be found. I propose a set of criteria to define factive thought experiments precisely. I show how factive thought experiments can be successfully used in philosophical reasoning by giving examples, mainly from Wittgenstein's work. In this way, I argue, it is possible to naturalize thought experimentation.

Quantum Mechanics, the Chinese Room Experiment and the Limits of Understanding

Quantum Mechanics, the Chinese Room Experiment and the Limits of Understanding