Blue book sales at the Cal Student Store are up by 80 percent over the last two years.
That’s one of the data points that California senior editor Nathalia Alcantara surfaces in her feature on the ways AI is changing higher education (“Test Case“), both in terms of how professors teach and how students learn, and sometimes cheat.
AI makes cheating easy—and hard to detect. It also blurs the line between academic fraud and simply taking full advantage of a new tool, the way a math student would use a calculator, say, or a photographer would use Photoshop. A lot depends on context. For some instructors, in some classes, using AI at all is prohibited. In other classes, AI use is encouraged or even mandatory.
At Berkeley, there’s no university-wide policy governing the new technology, and professors are experimenting with policies of their own. That’s understandable. It’s also understandable that some students, already ambivalent about a technology that threatens to eliminate careers, feel they lack guidance and face a shifting set of standards.
Many of the professors Alcantara spoke with freely admitted to using AI, including to assist with research or analyze data. They feel they’re knowledgeable enough to wield it responsibly. For their students, however, they worry it’s too often a crutch.
To make sure students are actually learning and not just letting AI do the thinking for them, old methods of evaluation like oral and written exams (those blue books) have become one obvious, if onerous, solution. How strange it would be if the apogee of digital technology delivered us back to blank notebooks and pencils.
It’s nearly as strange that Victorian-era technology still powers the future. I’m talking about suboceanic cables, which today carry virtually all international data traffic, including emails, phone calls, and AI prompts—albeit in hair-width fiber optic strands, not twisted copper wires.
That remarkable fact was brought to my attention not long ago by Professor Nicole Starosielski, who has created a first-of-its-kind certificate program and minor at Berkeley called Global Digital Infrastructure. As Starosielski’s course drives home, and our illustrated feature (“The Astounding Life of an Email”) shows, our supposedly “wireless” communications and “cloud-based” networks are anything but.
Our lightspeed communications are no less amazing for it. AI chatbots, in particular, feel like magic. Black magic, I often think. And yet I confess I use them—specifically, the free model of Claude, Anthropic’s chatbot. I don’t use it to write for me, but as a sounding board—as a tool, not a crutch. But maybe I’m kidding myself.
Before publishing this note, I shared it with Claude and prefaced it, “No doubt you see the irony and/or hypocrisy of me running this past you.”
The response came in seconds. “The irony’s real and I’ll let it sit there rather than perform discomfort about it—you’re the one who has to square it, not me.”
I’m not sure I can. I’m not sure any of us can.
