Abstract
It is not unreasonable to argue that artificial intelligence (AI) fulfils the Shannon–Weaver model of communication. Yet, AI not only impacts on human-to-human communication, AI is also about machine-to-human communication as well as machine-to-machine communication as we have known ever since Facebook’s ‘Bob and Alice’ AI talk.
Written by numerous authors and edited by Peter Klimczak and Christer Petersen, the book provides a comprehensive and interdisciplinary analysis of AI’s prospects and limitations. The book shows how AI is used to communicate between people and from machines to people and reverse. It dives into AI’s social and cultural aspects, as well as how AI is changing the nature of communication and social interaction.
The book comprises 11 sections, starting with a preface and followed by Learning Algorithms; Transgressing the Boundaries; Limits and Prospects; Big Data and Small Data; AI and/as Risk; Ethical Concerns, Contemporary Cinematic Narratives; Trends in Explainable Artificial Intelligence; Machine Dreaming., and concludes with a list of many experts with in-depth knowledge of many facets of AI.
The preface includes the story of Blake Lemoine – a Google programmer who raised doubts about whether an AI speech bot called Language Model for Dialogue Applications (LaMDA) had achieved consciousness. This prepares the ground for delving into the field of AI, debating its possibilities, and tackling issues like biases and poor data quality.
Section one, ‘Learning Algorithms’, explains what networks are and examines learning algorithms through examples like Alpha Go, and AI-powered chess computers. Chatbots – a potential application area for natural language processing (NLP) – provide examples of how AI enables two-way communication between humans and robots. ‘Eliza's case’ (p. 13–17) demonstrates the ways AI may simulate empathy and comprehension in its responses. This section ends by contemplating ‘Norbert Wiener's perspectives’ (p. 39), who was taken aback by how today’s utilization of AI deviates from his concept of AI acquiring knowledge through competition with humans.
Section two, ‘Transgressing the Boundaries’, examines the level of deep learning in AI today. It highlights the lack of theoretical knowledge in numerous areas, leading to preventing the creation of useful algorithms, and creating dangerous situations when safety is at stake. The authors address how numerical techniques affect the ‘effectiveness of neural networks’ (p. 60). Bayesian neural networks illustrate uncertainties and emphasise the need for a better understanding of learning algorithms’ strengths and weaknesses.
Section three examines the bounds and possibilities of ethical problems in the context of law and society using ‘accident algorithms’ (p. 83) in autonomous driving as an example. Human and machine decisions about life and death demonstrate how public discourse, particularly ethical issues, can indirectly impact technological and legal advancement.
Section four discusses how data analysis is used to generate small data. AI technologies now enable smaller firms to give tailored solutions because they do not require massive amounts of data or expensive computer systems. Using hybrid AI systems that combine ‘black box’ (p. 115) and ‘expert knowledge’ (p. 134) components can help effectively address the challenge of managing small data, thus being more approachable for those who are not specialists in AI. The section also demonstrates ‘NLP’ (p. 117) in chatbots, including text recognition, language generation/comprehension and machine translation, whereas cognitive modelling describes models that mimic human behaviour and intelligence.
Section five investigates the ethical and social implications of AI-enabled communication, such as its ability to facilitate human-machine relationships, weigh the dangers and advantages, and address AI’s tendency to perpetuate biases and discrimination. AI, in particular, has the potential to alter the nature of communication and social interaction. Finally, Chatbots and other AI-powered communication tools are more than just ‘artificial neural networks’ (p. 117); they are meant to represent specific cultural norms and communicational assumptions.
Section sixth opens the AI world widely: how technological knowledge, expectations and ethical convictions influence people’s perceptions of AI assistances. Individuals have been changing their attitudes towards novel technology and AI assistance. By comparing current fears about AI helpers to earlier anxieties about technology like silent movies and television, the authors highlight the significance of ‘digital literacy’ (p. 163) and the ability to utilise technology efficiently and discuss cognitive dissonance. It illustrates the uncomfortable feelings humans get when AI does not perform as expected or conflicts with their beliefs and goals.
Section seven – Man-Machines – delves into how novels and movies depict phoney women. It discusses classic examples such as Fritz Lang’s film ‘Metropolis’ (p. 195) and figures such as Pandora and Galatea. These stories mainly revolve around men who create fake women and the difficult relationships they have with them. The section concentrates on concepts such as control, creativity and the natural mystery associated with these phoney ladies. It basically looks at how the subject of men making and dealing with artificial women has been a frequent motif throughout history and how it still occurs in recent films about AI.
Section eight is about AI’s effects on daily life and its potential to revolutionise industries like ‘cybersecurity’ (p. 223) and healthcare. The researchers are particularly concerned about racial, ethnic and national origin biases in AI systems, prompting demands for more accountability and transparency. They suggested an Explainable Artificial Intelligence (XAI) method to handle the worries. It enables people to comprehend and enhance algorithms while increasing user confidence in AI. The theoretical underpinnings of XAI are applied to compile research data and suggestions for incorporating routine activities, and in healthcare. These findings make it simpler for everyone.
Section nine – Machine Dreams – demonstrates the development of AI, its effects on society, and the connection between AI and human intellect. The researchers try to answer the question if people should accept difficulties related to AI. AI, nowadays, is present in many facets of our everyday lives, and people are likely to need to understand its functions, and what it implies. The section reveals the influence of logical thinking and technology on human perceptions of their own distinctiveness compared to robots. AI impacts human beings’ ability to control things and the way people view themselves and maintain their sense of uniqueness even as AI improves.
Section ten starts with a playful title – Let’s Fool That Stupid AI. It focuses on ‘adversarial attacks’ (p. 268) on text-processing AI systems. The methods lead AI systems to make poor decisions or assessments. It provides readers with lessons to take away from researching these techniques and their impact on text-processing AI by using a chess game to illustrate these restrictions. Even while AI is quite skilled at the game of chess, it is not perfect. ‘Leela Chess Zero (Lc0)’ (p. 279) played Stockfish 10 in a game, but Lc0 couldn’t win despite appearing to have the advantage. This occurred as a result of Lc0’s inability to anticipate the implications of its actions; it could only recognise patterns. This demonstrates how AI struggles to comprehend ideas and concepts from the actual world.
The book AI – Limits and Prospects of Artificial Intelligence stands out for its all-encompassing approach compared to other books on this subject. This book provides a comprehensive analysis of AI, addressing its technological, ethical and societal facets; how AI influences human decisions and its risk assessment. Unlike this current book, Charles Jennings’ Artificial Intelligence: The Rise of the Lightspeed Learners only focuses on highlighting unexpected developments in the realm of AI. Simon Chesterman’s We, the Robots?: Regulating Artificial Intelligence and the Limits of the Law takes a more targeted approach and goes further into the legal and regulatory issues associated with AI.
Yet this current book might have possible limitations. The meteoric rise of AI may lead the book not to delve into detail on some issues in the context of sustainability and may quickly become outdated. The reader’s level of understanding of AI will determine the acceptability of the book. The different writing quality and style within the book might reduce reader engagement. This book may not always keep up with the most recent ethical breakthroughs in AI, at times, it might sway towards theory rather than practical recommendations.
In conclusion, AI – Limits and Prospects of Artificial Intelligence offers a deep understanding of the AI world. AI is not a neutral communication instrument. Rather, AI is influenced by our society and culture, including our values and concepts about communication. The book shows the influence of beliefs, moral, social impacts and concepts on how AI’s communication systems are produced and utilised. It is a valuable resource for those investigating AI and communication. Although beginners in the AI field might find challenges in this book, it still aids readers in understanding AI and how AI affects society by tackling technical challenges, and ethical issues.