During this time, my closest friend was extremely busy with work, I couldn't afford therapy because I was struggling to even make rent, and I didn't have someone I could talk to.

I have OCD. One of my compulsions is "checking" and it often takes the form of googling. Googling a word I know just to check that the definition hasn't changed. Googling whether xyz is normal. Googling how to choose some relatively unimportant things.

I googled all kinds of things. I googled to find out if there's a word for intentionally letting someone hear you cry in hopes that they'll comfort you. (Not because I was doing that, but because I had an overwhelming urge to.) I googled odds of relationship survival after addiction. I googled how to tell if your partner has stopped loving you.

And Google ai overview answered every single question. And on mobile, when I was curled up in the dark at 4am, there's a convenient text box to ask the AI a follow-up question. And the AI says things like "that's a hard situation to be in" and "that must be painful to hear"

And I'm not a fucking idiot. I know perfectly well that AI is trained to be sycophantic and that talking to a fucking LLM about my relationship fears was a massive mistake.

I was also terrified, alone, heartbroken, and desperate.

I spent literally hours of time googling questions on AI mode. Holding conversations with the AI. I have stickers on my laptop and water bottle that say "AI Hater Club" and "generative AI? Nein danke" and I wish Sam altman would fall off the face off the earth.

And none of that mattered. I was desperate. I recognized that, and once things began to calm down (my doctor took me off of the narcolepsy med and put me on a medication for addiction called Naltrexone, I was able to start therapy again, my partner and I worked together and took steps to repair what was damaged) I stopped turning to it for an open shoulder.

But I realized a few days ago that I hadn't stopped using the ai entirely. I would Google "best [item] on a budget" and instead of reading a cnet link that reviewed multiple brands, instead of checking consumer reports, I expanded the AI overview and asked follow-up questions.

I'm currently finishing up an astrophysics internship for the summer. Usually when I don't know how to code something, I look at the documentation or find a post on Reddit or stack overflow to explain. Instead, I was asking Google overview for instructions on how to use sys.argv. At no point was I vibe coding, but I was basically asking Google ai to walk me through it like a tutor.

And then, last week, I realized that I no longer Google things the same way. I used to search like "pokemon sun and moon original release date" but lately I've been searching "what year did pokemon sun and moon come out?" Even my checking OCD has changed. I used to Google "n30i mask cleaning +seal +oily" and now I Google "I use a resmed n30i CPAP mask and it keeps losing its seal when I'm trying to sleep. My face gets kind of greasy when I have it on. I already clean it with CPAP wipes every day and run it under water to rinse, do I need a different kind of mask?"

A few days ago, the news about Hank Green using LLMs to find research papers dropped, and I saw his response on Reddit. I saw the way he talked about the dopamine hit he got from it. And something clicked.

I had to install a different OS launcher on my phone so I could change the search bar widget to Firefox search instead of Google, because that's the only way to ensure I don't get the ai overview when I Google something. It's actually really frustrating because I instinctively search in long rambling questions, which does not work well. But I'm having a hard time going back to how I was.

#thank you OP for sharing this story #i have been really bothered by some of the ways people are talking about hank green and this sums it up well #there's a lot of talk about how 'being busy / stressed / neurodivergent isn't an excuse' #but he's not using it as one #he's not like. expecting immediate forgiveness #he's saying that he's realized his relationship to AI tools and his work in general is unhealthy and he's going to rethink it #and it really bothers me how little sympathy i've seen some people have for that #you can absolutely be disappointed—we expect better from a public figure with his values #(and as he's said his loved ones also expect better from him and so does he) #but we can't treat people who confess to having an unhealthy relationship with AI the same as AI sycophants #those are not even slightly the same thing #OP i'm sorry you've dealt with addictive AI behavior and i'm glad you're doing better now #and that Hank Green's situation has helped you recognize and mitigate the patterns you were still stuck in #i'm sorry for how the rhetoric about him must feel like it reflects on how people think of you #as a fellow AI hater we AI haters need to get rid of this idea that using AI ever makes you tainted and morally suspect forever #it helps no one and does so much harm