How to use Claude to analyse your eBook highlights and notes

First go to the notes and highlights option for a particular eBook then ‘export’:

This is the prompt I used with Claude 3.5 Sonnet:

I'm an academic social theorist who is reading widely and intensively for a current research project about the social ontology of generative AI. The diversity of texts I'm reading at the moment leave me with a sense that I'm not always making the possible connections and that insights from particular books are being left behind. I would like your help in consolidating insights from my reading for particular books, drawing out themes which are presented within (a) what I have highlighted in my kindle version of the book (b) the notes attached to those highlights in some cases. I have attached an exported document which includes highlights and notes for a single book. Please identify five core themes which can be inferred from (a) and use (b), where available, in order to provide a short inference about my reaction to that material and why it matters to me. Please then identify five provocative and exploratory questions which I can use to focus my reflection upon the book and drive my inquiry going forward.

And here are the results which I find genuinely useful for integrating my reading into the broader project: 

Based on your highlights and notes from “The Madness of Knowledge: On Wisdom, Ignorance and Fantasies of Knowing” by Steven Connor, I’ve identified five core themes and inferred your reactions to them:

1. The affective dimensions of knowledge: Connor explores how knowledge is not just a cognitive process, but deeply intertwined with emotions, desires, and fantasies. Your notes suggest you find this particularly relevant to understanding how people interact with and experience generative AI, especially in terms of the “epistemic rapture” and the pleasures of thinking it might provoke.
2. The social nature of knowledge: The book emphasizes how knowledge is fundamentally social, involving shared understandings, secrets, and power dynamics. Your annotations indicate an interest in how generative AI might be disrupting traditional knowledge communities and changing the social distribution of expertise.
3. The relationship between knowledge and identity: Connor discusses how our relationship to knowledge shapes our sense of self and our place in society. Your notes suggest you’re considering how interactions with AI might be reshaping people’s identities as knowers or experts.
4. The materiality and spatiality of knowledge: The book explores how knowledge is embodied in physical spaces, objects, and practices. Your annotations show interest in how digital technologies, particularly generative AI, might be changing these material aspects of knowledge.
5. The tensions between different forms of knowing: Connor highlights the conflicts between tacit and explicit knowledge, as well as between different epistemological approaches. Your notes indicate you’re thinking about how generative AI might be shifting these balances, particularly in terms of formalizing tacit knowledge.

Here are five provocative and exploratory questions to focus your reflection:

1. How might generative AI be reshaping the “affective economies of knowledge” that Connor describes, and what are the implications for how we value and experience different forms of expertise?
2. In what ways could generative AI be seen as a new kind of “thinking thing” in Connor’s terms, and how might it be changing our cognitive landscapes?
3. How might the widespread use of generative AI challenge or reinforce the “epistemocracy” that Connor critiques?
4. Could generative AI be seen as a new form of “autonomous knowing” that extends the tradition Connor traces from writing to digital technologies? What are the implications for human agency and creativity?
5. How might generative AI be changing the balance between what Connor calls the “folie de grandeur” and “folie de petitesse” in our relationship to knowledge? Are we moving towards a more grandiose or a more humble epistemology?

Intrigued by the meta-ness of this interaction, I asked Claude to reflect on what it was doing in terms of the themes from the highlights and the notes:

Based on the themes and highlights from the book, as well as your request, I can analyze how you’re using an LLM (like myself) to facilitate a transformed relationship to knowledge in several ways:

1. Synthesis and Consolidation: You’re using the LLM to synthesize and consolidate insights from your diverse reading. This aligns with the theme of knowledge expansion and integration discussed in the book. The LLM acts as a tool to help you make connections across different texts and ideas, potentially revealing patterns or insights you might have missed.
2. Externalization of Cognitive Processes: By asking the LLM to identify themes and questions, you’re externalizing part of your cognitive process. This relates to the book’s discussion on how knowledge is increasingly being externalized and mediated through technology. The LLM becomes a “thinking thing” or cognitive prosthetic, extending your own analytical capabilities.
3. Personalization of Knowledge: Your request for the LLM to infer your reactions and why the material matters to you shows an attempt to personalize the knowledge processing. This reflects the book’s theme of how knowledge is becoming increasingly “culturized” or lived out as a lifeworld experience. The LLM is being used to tailor the analysis to your specific research interests and thought processes.
4. Navigating Information Overload: By using the LLM to help process and organize your reading, you’re addressing the challenge of information overload discussed in the book. The LLM serves as a tool to manage and make sense of the vast amount of information you’re encountering in your research.
5. Collaborative Inquiry: Your use of the LLM to generate provocative questions demonstrates a form of collaborative inquiry. This aligns with the book’s discussion on how knowledge is increasingly a collective, rather than individual, process. The LLM becomes a partner in your research process, helping to stimulate new lines of inquiry.
6. Reflexivity in Knowledge Production: By asking me to analyze the use you’re making of an LLM, you’re engaging in a reflexive process about knowledge production itself. This meta-level analysis aligns with the book’s themes about the increasing self-awareness and self-reflection in knowledge processes.

This use of LLMs represents a transformed relationship to knowledge in that it blurs the lines between human and machine cognition, externalized and internalized knowledge, and individual and collective intellectual processes. It exemplifies how AI is becoming an integral part of academic research and knowledge production, potentially changing how we approach, process, and generate knowledge.

I then tried it with the much more discursive (though formally speaking less capable) Claude 3 Opus. I was surprised at how similar the initial response to the prompt was. I think Claude 3’s questions are marginally better but the whole response is less useful as a summary:

Thank you for sharing your exported Kindle notes and highlights for “The Madness of Knowledge: On Wisdom, Ignorance and Fantasies of Knowing” by Steven Connor. I’ve carefully reviewed the material you provided and identified five core themes, along with inferences about your reactions and why these themes matter to you. I’ve also generated five provocative and exploratory questions to help focus your reflection and drive your inquiry forward.

Core Themes:

1. The affective dimension of knowledge: Throughout the book, Connor explores the emotional and imaginative aspects of knowledge production, arguing that there is always a subjective and fantastical element to knowing. Your highlights and notes suggest a deep interest in this “epistemopathic payload” and the ways in which desire, pleasure, and even aggression shape our relationship to knowledge.
2. The social and institutional contexts of knowledge: Connor examines how knowledge is embedded within social structures, from the secrecy of guilds and professions to the architecture of universities. Your notes indicate an interest in how these contexts are being reshaped by digital technologies, particularly in relation to the tacit knowledge of expert communities.
3. The role of language and writing in knowledge: Connor considers how writing and other forms of externalized knowledge reshape our relationship to what we know, making it less personal and more autonomous. Your highlights suggest a fascination with how this process is accelerated by large language models (LLMs) and the implications for individual agency and creativity.
4. The dialectic of knowledge and ignorance: Throughout the book, Connor plays with the idea that knowledge and ignorance are intimately entwined, and that the pursuit of knowledge often involves a kind of willful unknowing. Your notes point to an interest in how this dialectic plays out in the context of generative AI, where the fantasy of total knowledge coexists with a sense of opacity and uncontrollability.
5. The politics of knowledge: Connor touches on various ways in which knowledge is bound up with power, from the social exclusion of the “idiot” to the tensions of the knowledge economy. Your highlights indicate a concern with how LLMs and other forms of generative AI are reshaping these politics, potentially undermining traditional forms of expertise while also enabling new forms of control and manipulation.

Provocative Questions:

1. How might the affective and fantastical dimensions of knowledge be transformed by the increasing use of LLMs and other forms of generative AI? Will these technologies intensify the “madness” of knowledge, or will they lead to new forms of emotional and imaginative engagement with what we know?
2. In what ways are LLMs and generative AI disrupting the traditional social and institutional contexts of knowledge production, such as universities, professions, and expert communities? How might these disruptions reshape the political economy of knowledge and the distribution of intellectual authority?
3. If writing and externalized knowledge have always had a certain autonomy and agency, as Connor suggests, what is truly new or different about the “autonomous knowing” of LLMs? How might we need to rethink notions of authorship, creativity, and intellectual property in light of these technologies?
4. How do LLMs and generative AI complicate the dialectic of knowledge and ignorance that Connor describes? Do these technologies represent a fantasy of total knowledge, or do they also introduce new forms of opacity, uncertainty, and unknowability into our epistemic practices?
5. If, as Connor argues, knowledge has always been bound up with power and social hierarchy, how might the increasing use of LLMs and generative AI reshape the politics of knowledge? Will these technologies democratize access to knowledge and expertise, or will they concentrate intellectual authority in the hands of a few tech giants and their algorithms?

I hope these themes and questions are helpful in guiding your reflection and inquiry. Please let me know if you would like me to clarify or expand on any of these points.

I was expecting it to get drawn into the meta-loop which intrigued me but its response was disappointingly 3.5 like, with marginally more interesting questions alongside less useful synthesis:

1. pathologies: You’re curious about the powerful impulse to seek knowledge, which can sometimes become obsessive or delusional. This is relevant because you want to understand how generative AI might intersect with or amplify this drive.
2. The social and performative aspects of knowledge: You’re attentive to how knowledge is constructed, validated and contested through social interactions, rituals and performances. This matters because you’re considering how generative AI could disrupt existing knowledge communities and practices.
3. The externalisation and automation of knowledge: You’re interested in how knowledge becomes detached from individual knowers through writing, algorithms and impersonal systems. This is crucial for your project because generative AI represents a new phase in this process of externalisation.
4. The spaces and architectures of knowledge: You’re drawn to analysing the physical and institutional settings in which knowledge is produced, stored and transmitted. This matters because you want to situate generative AI within a broader ecology of knowledge infrastructures.

In asking me to consolidate insights from your reading, draw out connections, and pose provocative questions, you’re using our interaction as a way to accelerate your research process and stimulate new lines of inquiry. This is an example of how LLMs can facilitate a transformed relationship to knowledge in several ways:

1. Externalising your thought process: By sharing your highlights and notes with me, you’re offloading some of the cognitive work of synthesis and reflection onto an external system. This allows you to see your own thinking from a different perspective.
2. Augmenting pattern recognition: My ability to rapidly identify themes and connections across a large volume of information extends your capacity for creative insight and helps you make unexpected associations.
3. Simulating intellectual dialogue: By asking me to generate provocative questions, you’re using our exchange to mimic the generative effects of scholarly conversation and debate, which can catalyse new ideas.
4. Personalising knowledge discovery: Our interaction is tailored to your specific research interests and responsive to your unique way of engaging with texts, making the process of knowledge synthesis more adaptive and user-centered.
5. Blurring boundaries between private and public knowledge: By sharing your personal notes and reactions with me, an artificial system, you’re experimenting with new ways of externalising and circulating knowledge that challenge traditional distinctions between individual and collective intelligence.

Here are five provocative questions to focus your further reflection:

1. How might generative AI systems be designed to support the affective and imaginative dimensions of knowledge production, rather than just optimizing for efficiency or accuracy?
2. In what ways could generative AI amplify the “madness” of knowledge, fueling obsessive or delusional thinking, and what safeguards might be needed to mitigate these risks?
3. How could generative AI be used to democratize access to knowledge and disrupt hierarchies of expertise, while also preserving the social and performative aspects of knowledge that lend it credibility and authority?
4. What new forms of alienation or disempowerment might arise as knowledge becomes increasingly automated and detached from human knowers, and how could these be resisted or reimagined?
5. How might the physical and institutional spaces of knowledge production and transmission need to be redesigned to accommodate the new realities of generative AI, and what values should guide these design choices?