Bauer M. AI and Common Sense. Ambitions and Frictions 2024 torrent

Textbook in PDF format

Common sense is the endless frontier in the development of Artificial Intelligence (AI), but what exactly is common sense, can we replicate it in algorithmic form, and if we can – should we?
Bauer, Schiele and their contributors from a range of disciplines analyse the nature of common sense, and the consequent challenges of incorporating into Artificial Intelligence models. They look at different ways we might understand common sense and which of these ways are simulated within computer algorithms. These include sensory integration, self-evident truths, rhetorical common places, and mutuality and intentionality of actors within a moral community. How far are these possible features within and of machines? Approaching from a range of perspectives including Sociology, Political Science, Media and Culture, Psychology and Computer Science, the contributors lay out key questions, practical challenges and "common sense" concerns underlying the incorporation of common sense within Machine Learning algorithms for simulating intelligence, socialising robots, self-driving vehicles, personnel selection, reading, automatic text analysis, and text production.
The point of this book is to examine the claim made by the proposition “AI with CS”, where common sense is to be an adverbial attribute of machine intelligence: doing things common-sensically. The engineering endeavour has since bifurcated into tool AI for specific purposes, so-called “expert systems”, and into general-purpose Artificial Intelligence (GAI; (sometimes also AGI, artificial general intelligence)); and GAI would be a method that is applicable to all problem types and works efficiently for large and difficult instances while making very few assumptions. AI is a technical field that races ahead with much hype and noise; we offer some reflections. The purpose is not to review the state of the art of knowledge representation, neural networks, Deep Learning, LLM, chatbots, robots nor GAI, but to clarify the reference to “CS” as invoked in these contexts. We clarify what CS brings and might bring to this development. CS is clearly a challenge for AI, and AI is a challenge for CS. However, we must avoid the temptation of a circular logic of defining operationally an otherwise opaque CS: if AI does CS, CS is what AI does.
A valuable resource for students and scholars of Science–Technology–Society Studies, Sociologists, Psychologists, Media and Culture Studies, human–computer interaction with an interest in the post-human, and programmers tackling the contextual questions of Machine Learning