Google's AI Overviews as a Teaching Tool

A new feature of Google search presents opportunities for information literacy instructors.

Google's recent decision to begin including AI Overviews at the top of their search results presents library instructors with a strong metaliteracy teaching tool. These overviews pull information from a variety of sources and present the user with an artificial intelligence powered overview of the topic or query. As of 5/30/24, this appears above all other search results. Librarians will be relieved to know that the cited information is linked under the overview - and the user has the ability to "check the source" if they are so inclined. Google's Search executive Liz Reid recently said this is all about letting "Google ... do the Googling for you."

I am certainly not the first to wonder if that might significantly reduce traffic - add revenue - for websites and content creators whose work is populating the overviews, and there have been numerous embarrassing examples of these AI Overviews simply getting it wrong. I encourage you to search for "Google AI Overview Fails" on your social media platform of choice for some funny - and some disturbing - examples. Leaving these questions aside for a moment, Google's AI Overview also presents those teaching information literacy with a powerful teaching tool:

  • These overviews open conversations about the overlap between information literacy and emerging AI technologies and platforms.
  • Because these overviews draw on such a variety of sources, they are inroads to talking about intellectual property and the ethical use of other's work.
  • They give librarians a concrete way to discuss AI - something that many academic libraries are stepping up and tasking themselves with. 
  • They give us ways to help library users better understand how and why they are seeing what they see online.
  • Those interested in critical librarianship can discuss the issues related to add-revenue loss for content creators, the ethics of encouraging users to trust a model that hallucinates, or even the environmental impact that this will have.
  • These overviews can also be seen as powerful research tools - and something worth discussing with increasingly busy students who can should be empowered to use them ethically and efficiently.

I hope that it is obvious that there are many ways that these AI overviews can be incorporated into a lesson about how information lives, and how it is created online. I can see this connecting to the ACRL Framework for Information Literacy and seems to overlap easily with several of the threshold concepts academic librarians are encouraged to address in our work with students. I will elaborate on this in future posts, but for now I want to highlight a specific activity you can use to start asking students these questions.

Assumed Intent and Google Search: Beyond Confirmation Bias

Present students with the following images, or generate ones more appropriate to the course content.

Ask students to briefly jot down, 'What do you notice about the two images below?'

 

What we are seeing here is the assumed intent of the Google search algorithm. The keywords we use actually do matter(!) and will change the results that we are presented with. Google is designed to present users with results that answer our question - not provide the most accurate or most authoritative information. Google assumes to know what we want based on how we ask. This is not necessarily a good or bad thing, and in fact is a great opportunity to teach students how to be better internet users and better researchers.  The AI Overview could be used to highlight all of these ideas to students - and I imagine students working in groups to find similar examples that could be shared out with the whole group. Hopefully in the end, students can better understand how using a more neutral vocabulary will yield better results. This also parallels the importance of using a controlled vocabulary when searching disciplinary databases, so I can easily see this activity leading into a disciplinary research workshop. 

 

These AI overviews are useful to students - and if we are encouraging them to be metaliterate learners and creators, I think it is incumbent on us to empower them to use these tools effectively and ethically. There is also room for us to encourage students to be critical of these tools - as more and more will certainly proliferate. Metaliterate learners need to must learn to evaluate not just the information that they are encountering but the platforms they are using to find that information. Encouraging students to explore Google's AI Overviews can be a very tangible, accessible, and practical way to get students doing just that.

As far as the difficult questions all of this raises, I think that is also up to the next generation to begin answering. After all, it will be their media, their work, their faces, and their sweet sweet content that has fed and will continue to feed these large language models.