AI With Friction
As AI becomes ubiquitous and all-encompassing, active efforts against cognitive deskilling are essential to avoid economic, psychological, and even legal risks | Edition #259
👋 Hi everyone, Luiza Jarovsky, PhD, here. Welcome to the 259th edition of my newsletter, trusted by more than 87,900 subscribers worldwide.
As the old internet dies, polluted by low-quality, non-expert, AI-generated content, you can always find raw, pioneering, human-made thought leadership here. Thank you for helping me make this newsletter a leading publication in the field.
*If you have friends and family members who are interested in AI, it is a great time to gift them a yearly subscription. 🎁 Add your special message here.
🎓 This is how I can support your learning and upskilling journey in AI:
Join my AI Governance Training [or apply for a discounted seat]
Discover your next read in AI and beyond in our AI Book Club
Receive our job alerts for open roles in AI governance and privacy
Sign up for our Learning Center’s free educational resources
Empower your team and save 40% with a group subscription
👉 A special thanks to HoundDog.ai, this edition’s sponsor:
Replit, the AI app generation platform used by over 45M creators, has integrated the HoundDog.ai privacy code scanner to proactively check for potential privacy leaks. Try the privacy scanner in Replit or download it for free to generate a local markdown report with a sensitive data map, data flow visualizations, and data flow traces for detected privacy leaks.
*To support us and reach over 87,900 subscribers, become a sponsor.
AI With Friction
Modern-life conveniences often come with cognitive downsides, especially when they reduce direct engagement with the intellectual or creative processes underlying the tasks being performed.
Sometimes the cognitive downside is minimal, irrelevant, or too specific, and most people would not care about regressing or fully losing that niche skill.
However, when automating tools become ubiquitous or all-encompassing, as it seems to be increasingly the case with AI, active efforts against cognitive deskilling are necessary to avoid economic, psychological, and even legal risks.
-
Let me use Waze, the navigation app, as an example.
I remember driving before Waze. I could often geolocate myself well and was particularly good at memorizing routes before driving to a new place. If I got lost, I could check the city map I had in the car, although it was a slow and, in some places, dangerous process.
As soon as I started using Waze, it became clear that automated routing was much more practical, and I became fully dependent on it. But because I stopped practicing, it felt as if my brain had turned off that skill.
City maps were no longer necessary, and I could go much farther without memorizing any routes. But because I no longer seemed to know how to geolocate myself, having the app crash or losing internet access mid-drive became a nightmare, especially at night or on unfamiliar routes.
For me, losing my ability to geolocate and memorize routes is not very important, so I accept the trade-off. I still use Waze and benefit from automated routing, knowing that I might never recover the geolocation skills I once had.
This is my conscious choice, and it is relevant for my personal case, my occupation, and my interests.
-
However, regarding generative AI tools that help brainstorm, draft, edit, and repurpose text, my choice is different.
I consciously choose not to use AI to write. Why?
I have been actively developing my writing skills since I was very young, and the process of expressing my ideas through writing has been intimately connected to the formation of my identity and my professional path.
I would absolutely not risk regressing or impairing this skill and losing all the intellectual, creative, and psychological benefits that I derive from the process of practicing this skill.
This is my specific case, and others will likely have a different relationship with writing and might choose to accept the trade-offs involved in giving up on practicing it.
I used these two examples to highlight cognitive deskilling associated with AI use. A way to avoid this, which most people are not aware of, is by intentionally using AI with friction: