After spending a week in Las Vegas for “Hacker Summer Camp”, I think I’ve figured out why some people become addicted to AI: it’s the dopamine loop from randomized reinforcement.
This dopamine loop occurs as a byproduct of how Large Language Models (LLMs) function. Probabilistic outcomes are a central mechanism for how the technology works – meaning that the occasional “hallucination” is actually a feature.
That same dopamine loop is what causes the feeling of maybe the next prompt will finally solve my problem!
Just like social media, slot machines, Gacha games, and other mediums that take advantage of randomized reinforcement to generate addictive behavior – I speculate that people become addicted to AI due to the built-in dopamine loop of “prompting something that magically works” (sometimes). I’m definitely eager to read more of the research happening in this space.
💭 A few quick thoughts
Check-in with your friends; the small kindness of simply messaging a friend you haven’t heard from in a while is likely to bring joy to you both. Making friends as an adult can be really challenging, and it’s important to remember that those hard-won friendships need to be rekindled from time to time.
Plan ahead, and be ready to adapt; being prepared reduces stress and allows us to enjoy being “present in the moment”. That said, being ready to adapt to the circumstances you find yourself in allows one to move fluidly through life. Very little ever goes exactly according to plan.
Pay it forward; we all have something to offer in terms of hard-earned wisdom from experience. When you shepherd someone through a challenging event or experience in their life, they will likely remember it forever – even if you might not. But, in the moment, you will experience the joy that comes from helping others. Pay it forward.
📚 From the bookshelf
The Shallows: What the Internet is Doing to Our Brains by Nicholas Carr. “Every information technology carries an intellectual ethic―a set of assumptions about the nature of knowledge and intelligence.” Published in 2011, it definitely feels like we’re overdue for a sequel in light of all that has happened over the last 14 years – and especially what has happened in just the last few.
Mastermind: How to Think Like Sherlock Holmes by Maria Konnikova. I read this book over a decade ago, and it was a far more enjoyable read than “Thinking, Fast and Slow” by Daniel Kahneman. It covers similar concepts, including systems thinking concepts and practical applications. I highly recommend giving it a read if you enjoy being fully present and observant.
Futureproof by Kevin Roose. “Whether you think AI and automation will be great or terrible for humanity, it’s important to remember that none of this is predetermined. Executives, not algorithms, decide whether to replace human workers.” Ah, yes – Capitalism.
📖 Recently read
Securing Agentic AI: How Semantic Prompt Injections Bypass AI Guardrails. Using emoji text as a prompt injection for generalized LLMs feels pretty novel (and fun!) ✍️ 🐍, 🖨️ 👋 🌎.
Lower Artificial Intelligence Literacy Predicts Greater AI Receptivity. A study found that people who don’t understand the way LLM technology functions experience a sense of “awe” when using it; they describe LLMs as “magical”. Indeed, any sufficiently advanced technology is indistinguishable from magic. Although, regrettably, LLMs are still just engineering.
Every Reason Why I Hate AI and You Should Too. “if they could actually think, with as much information as they have, there’d be groundbreaking discoveries literally falling from the sky.” While I have some disagreements with the post, Marcus Hutchins sort-of has a point 🤔 Why don’t we have a plethora of viable room temperature superconductors yet?
🤔 This week’s question
Once you’ve realized
you’re in a dopamine loop,
how do you break out of it?
Much of the software we use every day has been carefully engineered to generate and sustain dopamine loops. Breaking out of the loop usually requires creating friction between yourself and the software, service, or device that is creating the loop.
This could involve changing your phone / device to “grayscale” mode to reduce visual stimulation; adding a time limit for the application and/or website; uninstalling the software and only using it through a browser; or adding a “speed bump” with software like one sec.
What tips and tricks have you used to break out of dopamine loops?
Let me know what you think.
– Keith