Biology in AI: New Frontiers in Hardware, Software, and Wetware Modeling of Cognition
Artificial Life · 2023
DOI: 10.1162/artl_e_00412 ↗Kurz gefasst
Editorial in der Zeitschrift Artificial Life, 2023. Verfasst von Luisa Damiano und 1 weiteren Autor:innen. Beteiligte Einrichtung: IULM University, Research Group on the Epistemology of the Sciences of the Artificial, Department of Communication, Arts, and Media. luisa.damiano@iu…. Hauptthemen laut MEDLINE: Software, Computers. Die Arbeit nennt 3 Literaturangaben.
Abstract (Originalsprache)
The proposal for this special issue was inspired by the main themes around which we organize a series of satellite workshops at Artificial Life conferences (including some of the latest European Conferences on Artificial Life), the title of which is “SB-AI: What can Synthetic Biology (SB) offer to Artificial Intelligence (AI)?” The workshop themes are part of a larger scenario in which we are interested and which we intend to develop. This scenario includes the entire taxonomy of new research frontiers generated within AI, based on the construction and experimental exploration of software, hardware, wetware, and mixed synthetic models to deepen the scientific understanding of biological cognition.Since the early 1990s, basic research in AI, broadly conceived as the study of cognition and intelligence through artificial models, has been developed based on an increasing interest in insights provided by the sciences of life on natural cognitive structures, mechanisms, and processes. The growing importance of biology in AI research relies on recent paradigmatic transformations within this field, which, being enhanced by scientific and technological progress, are generating new research frameworks and directions.The emergence of the “embodied approach” in the cognitive sciences and AI, beginning in the late 1980s, has been increasingly focusing the investigations of this field on the role(s) played by the body’s self-production and interactions in cognition. As we argued in previous works on SB-AI (Damiano & Stano, 2018, 2021) this reintegration of the biological body, directed toward overcoming the Cartesian dichotomy characterizing the classical cognitive sciences and AI, appears to be only partially accomplished through the combination of synthetic hardware and software models—which are traditional in AI—and can be enriched by new approaches, based on the emerging wetware form of the synthetic modeling. Until now the latter has not received sufficient attention as a way of expanding AI research from the embodied perspective. On this basis, a potentially impactful scenario can be developed, through the complementary, critical integration—at both the experimental and the conceptual levels—of the three branches of the synthetic modeling mentioned, that is, hardware, software, and wetware synthetic modeling of life and cognition.By wetware models of biological and cognitive processes we refer to chemical models developed by SB, the nascent field constructing and studying biochemical artificial systems that display features of (minimal) life and (minimal) cognition. Whereas industrial SB aims at constructing parts, systems and/or processes to modify extant biological cells for practical applications, knowledge-oriented SB focuses on constructing artificial systems of various types, to generate, from the bottom up, living and cognitive patterns. As we emphasized in the works we dedicated to SB-AI, the constitution of this new field offers AI a new way of exploring the role the body plays in cognitive processes and converges with computer science and robotics in developing frontier programs in AI (in particular, in embodied AI) aiming at extending its research within the domain of synthetic experimental biology.The theme of this special issue, “Biology in AI: New Frontiers in Hardware, Software, and Wetware Modeling of Cognition,” can be conceived as collecting frontier research topics introduced to stimulate interdisciplinary avant-garde programs in AI involving the whole spectrum of the (hardware, software, wetware, and mixed) synthetic modeling of life and cognition. Some of the most fascinating and potentially productive research subjects connected to the theme of this special issue, are defined by the following questions:Considering SB’s current state of development, can it productively explore cognition and intelligence through the construction and experimental exploration of synthetic biological systems and processes? If yes, how? Under which conditions could this exploration positively contribute to AI research? That is, what can SB offer to AI?What are the groundings, procedures, expected results and impacts of current (or currently realizable) research programs involving SB in AI research? What novelties are they introducing into both the classical and the embodied approach to AI? What are its possibilities of enriching the traditional hardware and software synthetic exploration of cognitive processes?Can we, at the present time, plan concrete collaborations between computer science, robotics, and SB in the scientific study of natural forms of cognition and intelligence? How?What aspects of life and biological cognition can hardware, software, and wetware models, in their specificities and complementary combinations, allow us to explore scientifically and to better understand?What kind of synthetic models can be used to investigate, by means of specific models, relevant aspects of life and cognition genuinely emerging “from the bottom”?Are there innovative synthetic models for investigating (minimal) living and (minimal) cognitive processes, capable of drawing the lines of new theories or experimental approaches for the scientific characterization of life and cognition?Are there approaches able to trigger the transition from purely imitative AI to a form of AI focusing not only on behaviors, movements, and anatomic structures of biological systems but also on their organization?The articles selected for this special issue attempt to offer answers to some of these questions and to develop research directions for “Biology in AI.”The exploration of computational capabilities of chemical reaction networks is a research arena that combines AI and purely chemical systems well and aims at embodying computation into chemical transformations. Its implications are increasingly relevant when minimal forms of life are considered, for example, forms of life that potentially can be built by SB, to plan the construction of responsive and adaptive cognitive systems via a chemical perspective. In his article, Matthew R. Lakin presents an abstract chemical reaction network model that implements the backpropagation learning algorithm for a feedforward neural network whose nodes employ the nonlinear “leaky rectified linear unit” transfer function. The network directly implements the mathematics behind this well-studied learning algorithm, and the author demonstrates that it is possible to train the system to learn a linearly inseparable decision surface, specifically, the XOR logic function. The article, “Design and Simulation of a Multilayer Chemical Neural Network That Learns via Backpropagation” provides insights into chemical reaction networks as wetware machines capable of learning. While experimental biochemical strategies are called to action, and consider these targets as very relevant, yet challenging, research questions, computational approaches like the one developed in this article pave the way to focused projects and ultimately support embodied AI progressions.The article by Emanuela Del Dottore and Barbara Mazzolai, “Perspectives on Computation in Plants,” focuses on an emerging branch of biorobotics that applies the cybernetic paradigm of transdisciplinary circuits to the project of establishing a bidirectional exchange of knowledge between the scientific study of plants and the robotic construction of artifacts. On one hand, the research direction the authors present is engaged in developing robotic systems that model plant processes of morphological and behavioral adaptation to effectively perform in unstructured scenarios. On the other hand, this research line uses the robotic systems it develops as scientific tools, that is, material models whose experimental exploration can deepen the scientific understanding of plants. The ambition is to create a plant science that, based on this generative circuit of exchanges centered on plants, is able to promote both original approaches for preservation and management plans and innovative developments in science and engineering. The article offers a paradigmatic example of this kind of circuit by describing the role that the scientific exploration of functional behaviors of plants—in particular, their growth and communication behaviors—can play in developing algorithmic solutions for bioinspired robot controllers and, furthermore, in generating innovative approaches to robotic control architectures and networking.The article “Understanding Social Robots: Attribution of Intentional Agency to Artificial and Biological Bodies” by Tom Ziemke offers an interesting contribution to the philosophical dimension of embodied AI and Artificial Life by focusing on the specific problem of intentionality posed by socially interactive autonomous systems, for example, (some) social robots, automated vehicles, and behavioral robotized objects. With regard to these systems, Ziemke addresses what he calls the issue of the “dual nature of (attributed) artificial intentionality,” or the challenge of intentionality that they impose on their human users. On one hand, these systems require their users to interpret them as intentional agents to understand and anticipate their behavior, but on the other hand, to effectively interact with them, their users have to remember that these agents’ intentionality, being grounded in artificial bodies, is fundamentally different from human intentionality. Ziemke’s theoretical development of this issue generates a compelling invitation for the community of Artificial Life that proposes to develop the “practice of autonomous systems,” originally brought forward by Francisco Varela and Paul Bourgine (1992) as the core of Artificial Life, by introducing of a “practice of socially interactive autonomous systems.” This new practice is oriented by a deeper scientific understanding of the differences between artificial and biological bodies and the implications that these differences entail in the context of human–technology social interactions.In the final article, “Explorative Synthetic Biology in AI: Criteria of Relevance and a Taxonomy for Synthetic Models of Living and Cognitive Processes,” we address the topic of Biology in AI in two ways. First, we tackle the issue of the relevance of hardware, software, and wetware models, broadly construed as the problem of determining the concrete contribution that these synthetic models can make to the scientific understanding of life and cognition. Elements from cybernetic epistemologies and the epistemological framework of autopoiesis are used to introduce two assessment criteria for synthetic models, whose combination generates a wide, articulated space of possible forms of relevance for these models, located between the terms of the traditional alternative mere imitation vs. full reproduction of the target processes. In the article this space is defined through a detailed taxonomy of different forms of relevance that synthetic (hardware, software, and wetware) models can express in exploring biological and cognitive processes, with particular attention to biological cognition. Second, on the basis of this framework, we propose a tentative map of the forms of relevance characterizing wetware models of living and cognitive processes produced by SB today and define a programmatic direction for the development of “organizationally relevant approaches” applying SB techniques to the field of (embodied) AI.Through these articles, the special issue Biology in AI promotes original directions on AI within and through the scenario of the synthetic modeling of life and cognition. The hope is that, as stepping stones, the articles that we collected in this special issue can inspire and influence novel ideas, debates, collaborations, studies, and research domains across the scientific communities converging into Artificial Life.In addition to the themed articles discussed, this special issue includes the third installment of Chris Adami’s column “The Evolutionary Path to Sentient Machines.” The overall topic is on evolving intelligence. In this installment, Adami discusses five elements of intelligence in terms of information and its acquisition, storage, and manipulation.We acknowledge the productive collaboration of the Artificial Life co-editors-in-chief Susan Stepney and Alan Dorin, and the precious, patient, and supportive help of Lin Reedijk of the editorial office. This is also an opportunity to thank the authors and reviewers of the published articles, as well as all contributors we have hosted as speakers in the SB-AI workshop series over the years. The scientific collaboration, as workshop co-organizer, of Yutetsu Kuruma (Japan Agency for Marine-Earth Science and Technology, Yokosuka) and, as critical commenter and discussant, of Hagen Lehmann (University of Bergamo) is also greatly appreciated.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- IULM University, Research Group on the Epistemology of the Sciences of the Artificial, Department of Communication, Arts, and Media. luisa.damiano@iulm.it.
- University of Salento, Department of Biological and Environmental Sciences and Technologies. pasquale.stano@unisalento.it.
Zitieren
APA
Damiano, L., & Stano, P. (2023). Biology in AI: New Frontiers in Hardware, Software, and Wetware Modeling of Cognition. Artificial Life. https://doi.org/10.1162/artl_e_00412
Vancouver
Damiano L, Stano P. Biology in AI: New Frontiers in Hardware, Software, and Wetware Modeling of Cognition. Artificial Life. 2023. doi:10.1162/artl_e_00412
BibTeX
@article{Damiano2023,
author = {Damiano, Luisa and Stano, Pasquale},
title = {Biology in AI: New Frontiers in Hardware, Software, and Wetware Modeling of Cognition},
journal = {Artificial Life},
year = {2023},
doi = {10.1162/artl_e_00412}
}
EUVIMED