“We’ve been the most trusted name in research for over 150 years” says Nature on its “Nature Research Intelligence” site. You’d think they’d be a little more careful with that reputation.

Promising “insights you can trust,” Nature Research Intelligence offers around 30,000 “topic summaries” across multiple fields, each “created with generative AI and the cited references.”

The results, at least in philosophy, are… not good.

First, for no stated reason, they lump together philosophy and religious studies.

Second, the summary of recent research in the field is a bizarre, random sliver. If you happen upon this page looking to learn what philosophers are researching, here is the overarching picture Nature Research Intelligence offers:

The fields of philosophy and religious studies explore fundamental questions about existence, knowledge, ethics, and the nature of belief systems. Recent research in these areas has delved into various topics, including the nature of consciousness, the psychological underpinnings of conspiracy beliefs, and the intersection of religion and politics. These studies not only enhance our understanding of human thought and behavior but also highlight the complexities of belief systems in contemporary society.

Each entry offers definitions of key “technical terms”. For the “philosophy and religious studies” entry, three such terms are defined: “consciousness”, “conspiracy beliefs”, and “White Christian Nationalism”—basically all the concepts you need to get caught up in philosophy and/or religious studies.

Within the Philosophy and Religious Studies category is a distinct entry just for “Philosophy“. It’s not much better and still quite odd, mentioning only the following four topics as recent research foci within the field: “consciousness”,”mental disorders, particularly through the lens of lived experiences,” “active inference”, and in case you missed it the first time, “the relationship between subjective experiences and mental disorders.”

Things get slightly less weird as you drill down, but the topics and entries are still oddly idiosyncratic, narrow, and repetitive. Here’s what it says are the top three “significant areas” of “recent research” in social and political philosophy: “hope in the face of global challenges,” “the idea of community and its importance in fostering solidarity,” and “the ‘Focus Theory of Hope'”.

Here’s the page for various entries on topics in ethical theory. Now I appreciate fine-grained distinctions as much as the next philosopher, but this category includes entries on both “Metaethics and Moral Philosophy” as well as “Moral Philosophy and Metaethics“—and these are, according to the descriptions Nature Research Intelligence provides, different research areas.

In case you didn’t believe me

Nature says on each entry at Nature Research Intelligence that:

We take care to ground generative text with facts, and have systems in place to gain human feedback on the overall quality of the process. We however cannot guarantee the accuracy of every summary and welcome feedback.

But this disclaimer is rather weak sauce, coming as it does from a project associated with the prestigious “Nature” name. I doubt any human being, let alone anyone who knows about philosophy, actually did any kind of quality check on its material. They don’t even bother to classify the project as being in “beta”.

It’s not that using AI for projects like this is necessarily bad. But at this stage of the technology, the absence of human attention and curation has produced garbage. This kind of rush by knowledge-producing and knowledge-disseminating institutions to make use of AI, be it to increase their web traffic, save money on labor costs, show they’re on board with the next big thing, or whatever, is irresponsible. And it’s part of the enshittification of the internet.

Since 2015 Nature has been owned by Springer (Springer Nature). So perhaps if social responsibility isn’t one of their chief worries, the threat to their reputation (and hence the value of their gatekeeping, and hence their bottom line) may motivate them take it down and re-release it when they can do more about its quality than give us a virtual shrug.