Philosophy Cannot Be Read Quickly

Google recently announced a new feature for Google Scholar, the part of their search engine that searches through academic articles and books. Now you can open a “Quick Read” pane for select papers that includes an AI-generated summary, along with information about the approach or methods used in those papers, as well as “considerations” in the form of “caveats to keep in mind as you analyze the results.” The purpose of a “quick read” is ostensibly to speed up research: instead of reading entire long and technical papers, you can get answers to your research questions much faster by skimming AI summaries.

“Quick reads” are available not only for scientific papers, but also for works of philosophy. Curious, I decided to put them to the test with the works of philosophy I know best: those that I have written. In a test both of the accuracy of Google’s AI and the limits of my own narcissism, I performed some Google Scholar searches on questions or topics I figured had a good chance of bringing up my own papers to see how useful these “quick reads” really were.

Overall, the high-level overviews the AI provided were a decent summary of what the papers were generally about. After that, the results were discouraging. Specific philosophical arguments I made, or those I summarized from other authors, were often mischaracterized. The argumentative structures of the papers were absent, and any subtlety or nuance in points I had made was missing entirely. Perhaps worst of all were the “considerations,” which either simply reiterated limits I had presented in laying out the scope of the paper (and so not really a “consideration,” per se), or outright non-sequiturs (in one instance, warning the reader that I had not provided my full contact information, and so apparently was not to be trusted).

Aside from the inaccuracies and considerations that aren’t really considerations, what feels truly offensive about a “quick read” when it comes to works of philosophy is that philosophy is not meant to be read quickly. Indeed, in a world where AI tools are encouraging speed above all else, philosophy reminds us that meaningful engagement requires us to slow down.

Lots has been said about the detriments of living in a world of constant distractions. Attempting to fill all our own downtime may adversely affect our ability to consolidate memories. So-called “digital overload” – the constant stream of messages, alerts, and notifications we receive from digital devices – impairs our ability to focus, manage tasks, and remember what we’re doing. AI risks making these problems worse: while ostensibly meant to take mundane tasks off your plate so that you can focus on what’s important, some researchers are concerned that reliance on AI tools can erode introspection, create an over-reliance on feedback from AI, and induce anxiety due to worrying about not optimizing your work enough. At the same time, there are also concerns that the relentless pace of the digital world has fried people’s attention spans to the point that we aren’t reading books anymore, a problem especially emphasized by teachers.

These are all complex problems with no one-size-fits-all solution. But perhaps engaging with philosophy can help. Works of philosophy (be they your classics, modern journal articles, or works of public philosophy like you’re reading right now) demand a particular kind of attention. Since they present structured arguments in support of conclusions, philosophical works cannot be engaged with passively: reading philosophy requires you to examine your own beliefs, challenge assumptions you may have taken for granted, and, if you find yourself disagreeing with the author, reflect on why that might be. If an AI-infused digital world reduces our opportunities for sustained attention and introspection by offering quick, easily-digestible answers, then philosophy can help keep our intellectual muscles from atrophying.

For these reasons, among others, efficiency should not be the goal of engaging with philosophy. But let’s say you want to dig into a philosophical issue, but don’t want to spend time reading a bunch of different books or papers, running the risk of wasting your time reading something that isn’t what you’re looking for. It is the age of AI, after all; isn’t it about time someone automated the tedious aspects of research?

Here, we might again worry about inaccuracies contained in AI summaries. In my own tests (with admittedly a very small sample size), I found numerous errors, both small and large. When it comes to Google’s AI summaries in general, estimates are that they get things right 9 times out of 10, although these numbers are hard to verify. Depending on our tolerance for error, we might think this is either too high a rate of error, or within an acceptable range.

But accuracy rate is mostly beside the point. Sure, an accurate AI summary is better than an inaccurate one, but if you want to engage with philosophy through AI summaries, then you’ve already made a foundational mistake. This is not to accuse those reading AI summaries of “cheating”: philosophizing via AI isn’t taking a shortcut; it’s missing the point of philosophy altogether.

In his 1877 article titled “The Ethics of Belief,” philosopher W. K. Clifford argued that there are strict epistemic and moral requirements for forming beliefs, such that “it is wrong always, everywhere, and for anyone to believe on insufficient evidence.” Recognizing that he is being demanding, Clifford considers a discussion with a hypothetical interlocutor who doesn’t want to put in the work of ensuring that their beliefs are well-grounded in evidence:

“But,” says one, “I am a busy man; I have no time for the long course of study which would be necessary to make me in any degree a competent judge of certain questions, or even able to understand the nature of the arguments.”

Clifford’s response: “Then he should have no time to believe.” 150 years later, we can imagine a similar complaint coming from someone who, empowered with new tools and digital distractions to attend to, claims that they don’t have the time to put in the time or effort to engage with philosophical works, and that they wouldn’t know what to think about them if they did. Perhaps then they would turn to AI to summarize papers, analyze arguments, and generate “considerations” on their behalf. But while Clifford was arguably being overly critical, it doesn’t seem nearly as harsh to say that if you don’t have time to do philosophy slowly, then you should have no time to do it at all.