Reading with AI
Key Ideas
Artificial Intelligence, particularly the ready availability of AI summaries, creates a risk that knowledge makers might stay only on the surface of their understanding of what they read.
Reading can be enhanced if summaries or chatbots interactions are used to triangulate information: to develop a “conversation” between the person, the reading, and the chatbot. With the right approach, generative AI and AI summaries could immerse people in further in what they read and potentially lead to deeper understanding.
Many learning contexts do not pay enough attention to the ways reading and writing practices interrelate. Carillo (2017) warned that, without skills to engage meaningfully with sources, learners may take texts at face value. When students cannot read critically, they grasp ideas only partially
This is a problem that can be exacerbated by the presence of AI reading summaries. AI summaries can be useful for people who already have disciplinary knowledge. They are more fraught for people learning disciplinary ways of knowing and doing. There is a risk that AI summaries miss important dimensions of the research conversation with which a resource engages (Aharoni et al., 2023; Cao et al., 2022). Because AI tools tend to scan and articulate the broad, surface-level dimensions of a text, they miss nuances.
For Example...
If a student was scanning only for whether an article explicitly discussed a keyword or how prominently that keyword is discussed, there is a risk that a related resource could get overlooked because the language it used or the approach it employed (e.g., if the text used metaphors to describe concepts) did not align with the keyword or method of finding sources.
So how can AI summaries be used well? Familiarity with a text is the best starting point for making connections.
AI summaries are most useful when they add a layer to how one reads (Dang et al., 2022) and facilitate understanding disciplinary ideas (Watkins, 2024). This added layer can formulate connections that allow writers to act and contribute within their disciplines.
Generative AI can offer a rhetorical space for learners to cut through jargon, ask questions of the summary and the article (combined), and begin forming connections with the texts they read (Eaton, 2024). It is important in this scenario that the summaries are not the only means of procuring information, but it is conceivable that this is another triangulation strategy wherein learners can connect their own reading with both the AI summary and the dialogue they have with AI tools about the text.
With this knowledge, it becomes easier to think about how AI summaries can be applied to make reading more effective. The most beneficial reading practices emphasize the link between reading and writing.
Below are two ways that AI summaries can accomplish these goals:
AI Summaries for Triangulation
If learners read an article themselves first, they build an understanding of how the text deploys concepts and terminology; they pick up on context and the central claims being made (Eaton, forthcoming, c).
Summaries cannot stand in for reading or remove the requirement to return to the text. What they can do is strengthen understanding by giving readers another reference point to consider alongside other aspects, such as the abstract, keywords, and their prior knowledge of the material.
Remember
Reading and writing are interconnected dimensions of knowledge making processes (Sullivan et al., 2023). Readers join scholarly conversations and then contribute to those conversations (Carillo, 2017; Giordano et al., 2023)
Generative AI and AI summaries could stimulate a conversation around a text. With the right prompting generative AI could help people consider other dimensions of the texts they read. This only works, however, if people have read the text and understand its contents before engaging with the summary