How, When, and Why to Use AI

This is the fifth in the series Get Your Facts Straight: Research Skills for Writers. For more about this 18-part series, including the complete schedule and links to the other articles in the series, click here.

---

AI is not a substitute for research.

There are times when you may wish to use AI as a tool to enable a part of your research.

Why isn’t AI a substitute for research? Simply put, because it can be nowhere near as thorough, as creative, or as principled as you can.

Why might there be times when you might want to use AI as a tool to enable part of your research? Essentially, AI is useful for research in the same ways that a library catalog or other similar catalog database is useful for research: it provides a way for you to quickly and effectively assay a large pool of potential sources and get a sense for what’s out there that’s related to your topic. But there are important differences too. Issues that have outsized influence on how useful AI can be to research include information pool size, how much of a source is searchable, and source quality issues including references and provenance.

The world’s largest swimming pool is at the Sharm El Sheik Citystars resort in Egypt.

Information Pool Size: One major difference between a catalog database and an AI system is the size of the information pool they work with Catalog databases deal specifically with the documents and data that have been included in their specific database, for instance the items held by a particular library or consortium of libraries (let’s say WorldCat), or perhaps the content of all the content accessible under one particular content platform (like JSTOR). This means there are built-in limitations on what kinds of materials are discoverable when you’re using these tools: they index only what their specific databases include.

AI systems, by contrast, may range far wider than any given collection or set of collections. This is both a blessing and a curse. On the one hand, it means that AI systems may represent a broader array of the information out there (at least if it’s electronically accessible); on the other hand, it means that AI systems often do represent a broader array of the information that exists there, but may not do so usefully.