AI and reading tools
Artificial intelligence also enters into how we read, or perhaps better - into conceptualising how we can interact with text. It seems to largely redraw the concept of reading as a privileged way of working with data. It focuses on various shortcuts and simplifications that can be useful to us when working with text.
The key question these tools pose is what it means to read - what we want from working with the text and how we want to work with it. Haruki Murakami often works with the concept of "fast food" in his books; his fictional characters prepare a quick meal for hours. Food does not serve only to fill you up; this author rejects fast food because of the quality and, more importantly, the lack of relationships. To prepare a meal is to think about it, about oneself, to enter a time of some transformation. This is precisely the dimension we need to consider when working with AI in critical text work - we may get the information we need quickly and be able to transform the content, but we may lose something in the process.
In one of his books, Bruno Ferrero describes a dialogue between a woman who goes to church and a household member who does not. The man asks the woman what the priest said in the sermon. And she replies that she doesn't know, that she doesn't remember. When asked why she goes if she doesn't get anything out of it, the woman replies it's like the salad she washes for lunch. There's no water on it, but it's clean. And it can't be said that it doesn't give a person anything. The emphasis on the reading process is the same - sometimes, it's about the facts, the points, and the concepts, and AI can be a perfect helper. But other times, it's just about transforming or changing our thinking through reading. And then we have to read, then time is the tool that allows us to change.
The reference to the two writers is not accidental, but it is meant to create a specific space for critical reflection or thinking about what we will use the tools for. To read Franz Kafka's The Metamorphosis by reducing the text to one or two pages is to miss the text entirely. On the other hand, sometimes we need to get an idea of an article and its content because we need to get to some specific information or are looking for an answer. Then, on the other hand, turning through text that we scan with our eyes is not entirely formative. So, the key question is what we hope to get from reading.
At the same time, we must point out that an AI system may not be accurate and reliable and that AI, if it is to be productive and creative, will necessarily lead to the creation of errors and inaccurate answers that force humans to think about the problems and topics selected. It may impose a model of world structuring or wrong reasoning against which we can critically delimit ourselves. Still, this delimitation will also be part of the epistemic field with which we will continue to work. Reading with AI can bring speed, economy, and volume of texts analysed, but the fact that good journals all forbid the use of AI for analysis has good reasons. Indeed, working with it in this area is only appropriate, possible and thought-provoking in carefully considered areas.
It is not our goal to analyse in detail all the tools that can be used to support reading, nor is it possible - after all, they constantly appear and disappear, changing their functions, language models and prices. But we would like to look at least at a specific typology of tools that can be encountered in this field and with which one can work on the process of critical work with text:
Summary. A relatively large number of tools are dedicated to summarising and analysing academic texts. Examples include SciSpace Copilot, which can focus on topics related to your research, or Scholarly, which allows you to create summaries of the text, especially academic text, without registration but can also create flashcards for learning from the material. For general summarisation, you can then use, for example, Scribbr (with many additional features) or PDF Summarizer, which also has many annotation features.
Questions and answers. An interesting area is the tools that allow you to have questions over a file and then track down the place with the appropriate answer. Examples are ChatPDF or ai. They make it relatively easy to see if a document addresses what you need - i.e. you know it does when it hits, but if you need to be sure, you still have to go through it yourself.
Conversion to sound. For many people, it is practical to be able to listen to text as sound. Perhaps the most prominent tool in this area is Google's NotebookLM, which can do summaries, answer questions, or turn a text source into an audio discussion show. For converting text to audio in general, you can use, for example, Ultra Realistic Text To Audio or ReadLoudly.
Recommending additional content and working with books. Here, the choice is extensive. For example, the Heartbitz.app focuses on summarising and selecting interesting news in audio form. ai is used to recommend other books based on your previous reading and tries to be a kind of digital librarian, and Next Three Books works similarly, where you can choose the length or style of a book, for example. ChatfAI allows you to chat with literary characters.
Knowledge synthesis. In this category, Consensus is probably the best-known tool, to which it is possible to ask expert questions, and this system tries to offer an answer built from various referenced sources. However, references to other sources can now also be offered, for example, by ChatGPT.
