What to Read about Generative AI: Summer 2026 Edition

Four readings on AI's impact on the environment and education

Posted on July 6, 2026 in Blog Posts

Although both OpenAI and Anthropic, the companies behind ChatGPT and Claude, respectively, have filed for landmark IPOs later this year, the story that best encapsulates the state of AI discourse is Pope Leo’s recent encyclical, “Magnifica Humanitas.” Its subtitle captures much of its tenor: “On Safeguarding the Human Person in the Time of AI.” Popes have seldom shied from expressing their beliefs on current affairs, but Pope Leo’s popularity seems, in part, due to his willingness to confront AI, which polls very badly. Pope Leo is speaking to a mounting resistance to AI, which has yet to find political expression but is building in public opinion.

Indeed, the other telling set of anecdotes from the past month has come from graduation speeches at university campuses across the US. Rather than cite a sophomoric quote from a Steinbeck novel, several commencement speakers used their speeches to forcefully assert that AI would set the terms of the future. Students booed in response. Since the public release of ChatGPT, we have assumed that students would embrace AI as a partner-in-crime if nothing else, and universities have rushed to forge new majors, minors, concentrations, certificates, and stickers to mint students with the AI credentials they supposedly crave, and yet these same students—whose college educations were coeval with the rise of AI—seemed to reject it. 

I like to begin these periodic roundups of AI articles with such anecdotes to survey the intellectual landscape. I think we would be careless to forecast from these two stories a popular and total rejection of AI—a “hype” narrative in its own right. Rather, what they indicate to me is that the social, economic, and ethical costs of AI are moving into the fore. There is a desire to determine where AI belongs in human life (and where it doesn’t) and what norms and safeguards will police it. To that end, I have selected four readings from the reams of posts about AI that take on the messy work of answering these questions. 

---

“Environmental Impacts of Artificial Intelligence” (Ithaka S+R) 

The environmental costs of AI loom large in discussions surrounding its ethics and use. Companies are often vague about the environmental impact of their tools, and even if they were more transparent, institutions and individuals would likely remain in an ethical morass as they navigate the trade-offs. But inherent complexity is not a good reason to dismiss environmental concerns as murky, insignificant, or ultimately unknowable. 

Ithaka S+R has done us all a big favor by compiling a LibGuide on AI and the environment. Its multimedia offerings include general overviews of the issue, impacts on specific areas of the environment (e.g., water, emissions), ways to mitigate or respond to the environmental harms of AI, and further readings. It’s hard to overstate the richness of this LibGuide, which will provide librarians, students, and researchers with an important entry point to this big issue. 

---

🔥 Stay up-to-date with LibTech Insights by signing up for our free newsletter. Just one weekly email with our new blog posts, top tech news stories, and other bonus content. Check out some posts from our archive:

• Beyond the Job Title: Electronic Resources Librarian
• How Three Academic Libraries Merged Their e-Resource Collections
• Three Takeaways for Academic Libraries from the 2026 Library Systems Report

📣 Do you teach AI literacy? We want to hear from you. Check out our call for papers, “AI Literacy Across Disciplines.”

---

“Making Sense of GenAI Amidst AI Hype and AI Personalization,” by Sarah Morris (In the Library with the Lead Pipe) 

Morris’s article tackles two topics: AI hype and AI personalization. Although the former has received much attention, the latter, to my knowledge, has never been discussed in the literature on AI literacy, and yet Morris makes the case for why we must engage with it.  

“AI personalization” refers to the social techniques that AI chatbots deploy to build a warm rapport with users, such as flattery, agreeableness, and sycophancy. For instance, a user might express an idea to a chatbot, and the chatbot will praise them for the sophistication of their analysis or the genuine novelty of their insights—compliments that are hard to ignore and that naturally shape the user’s perception of the chatbot. Indeed, Morris argues that AI personalization creates trust and awe, encouraging users to drop their guards and engage with AI uncritically. 

Morris’s article is well worth the top-to-bottom read, but her best point is that AI personalization means that AI chatbots are designed to defuse critical thought. It’s difficult to be skeptical of someone who says that you’re right, and even harder if you impute that “someone” with superintelligence and perfect rationality. Students, by instinct, look for guidance and confirmation as they take their first tentative steps into academic research, and AI is more than willing to provide this reassurance. AI literacy needs to teach students to be critical of this dynamic.

“Dialogues with AI Reduce Beliefs in Misinformation but Build No Lasting Discernment Skills,” by Anku Rani et al. (CHI) 

MIT made headlines last year with a study that tested the neurological consequences of LLM-assisted essay writing. Its findings popularized the term “cognitive offloading” in pedagogical discussions of AI, referring to the reduction of cognitive activity that occurs when people use AI for intellectual tasks. 

A new study from MIT focuses specifically on identifying misinformation. This four-week study had 67 participants classify whether headlines were real or fake with assistance from an AI chatbot. The study found that participants, when assisted by AI, could identify misinformation with greater accuracy (+21 percent); however, afterward, they were significantly less able to identify misinformation without AI (-15 percent). What researchers concluded has implications for all educators—namely, AI can help with tasks but doesn’t seem to be a good teacher. AI assistance did not lead to skill acquisition. To quote the study: “Instead of developing independent discernment skills, participants increasingly relied on AI validation.”

When we consider educational policy, this discrepancy between short-term benefits and long-term harms is especially important to keep in mind: it’s easy to imagine a scenario in which policymakers and school administrators cite data showing that AI improves educational outcomes, but those numbers may not reflect actual, long-term learning. 

While this study may seem like additional evidence that we should eject all AI systems from the classroom, the researchers put forward another solution. Their takeaway is that chatbots need to be designed to cultivate discernment, perhaps through Socratic questioning. Such methods might prevent skill deterioration and instead foster active engagement.

“PEACEful Use of AI: A Tool for AI Education,” by Kimberly Shotick (ACRL) 

The goal of this roundup is to shine a spotlight on articles that are working through the messiness of AI, and the pedagogical framework Kimberly Shotick puts forward in this piece for ACRL offers one way for librarian-educators to engage students in this messiness. Shotick uses the acronym PEACE (Policies, Ethics, Agency, Critical thinking, and Environment) to sum up a series of critical questions meant to prompt a “critical pause” when using AI chatbots. This pause creates an occasion for students and researchers to think more deeply about AI systems and mitigate their harms. 

In this piece, Shotick outlines the critical questions that accompany each letter of the acronym. For instance, “Agency” asks users to consider whether they retain control over their data and its use. These Socratic questions are precisely what Rani et al. identified as absent in their study, so Shotick’s framework may offer a way out of this problem. Anyone actively engaged in teaching AI literacy should take a close look at the PEACE framework. 

And just a bit of a teaser—if you like Kimberly Shotick’s work on AI pedagogy in this article, you should register for LibTech Insights/Clarivate’s three-part micro-course on AI literacy, “Beyond the Basics.” Not to spoil anything, but you might be hearing more from Shotick in the unit on pedagogy. 

Related Posts