This book offers a critical reconstruction of the fundamental ideas and methods of artificial intelligence research. Through close attention to the metaphors of AI and their consequences for the field's patterns of success and failure, it argues for a reorientation of the field away from thought in the head and toward activity in the world. By considering computational ideas in a philosophical framework, the author eases critical dialogue between technology and the humanities and social sciences. AI can benefit from new understandings of human nature, and in return, it offers a powerful mode of investigation into the practicalities and consequences of physical realization.
For me, Agre is everything critical and constructive where I found Dreyfus insulting and unhelpful. I would have been impressed just by the community analysis that he does; the early chapters' explanation of the engineering mindset among AI researchers and why they respond to critiques the way they do was exactly the reasoning that I wanted to hear. But he goes beyond that explanation to actually make a critique of AI, and in actually demonstrating alternatives and working through their issues, I find his critique much more meaningful and persuasive. That is the kind of critique that I've always wanted the I School to be a place for.
Also, I love the programming language and that it's called Life. I couldn't easily find implementations of "Life" in Python or other modern programming languages, and I wonder if it's possible to build agents with it. And I'm curious reading these examples what it would be like to try to implement, say, a Slack chatbot that built up running arguments or used dependencies to learn new reasoning during conversation.
Partly, I find it hard to take critiques seriously when they're disconnected from practice. That Agre includes lines of code and step-by-step analyses makes it that much easier for me to accept that his position is informed by an understanding of the work, but more than that, I think the descriptions of technical detail (like the historical piece about logic gates and clock time) better explain the metaphors and assumptions behind computer engineering. The Pengi example is useful and detailed in its description (although I'm disappointed that I can't see it running somewhere), but I had optimistically hoped that there would be even more guidance and description on implementing the interactionist metaphor, and perhaps in a domain not as constrained as a video game.
Interesting to me for helping me think how to go about doing research itself. Will probably have to read it again. Content wise, I think modern AI research is already follows some of the proposed world views. Although more critical thinking may not be as widespread.