I fell into the trap of feeling left behind starting last fall through most of this year...until I also took a step back and realized this is now what I want out of my life.
Remember talk of the four day work week??? Why isn't AI helping us achieve THAT instead of having agents run 24/7 and pinging us for permission every 20 minutes?!
Great article, Deb! Longtime reader, first time commenter. I appreciate the call to action to allow ourselves grace with the AI transition and also echo your perspective that the feeling in SF around AI is unlike anywhere else.
Just one note - “AI psychosis” is a term psychologists are starting to use to describe delusions induced by chatbot interactions, not the general hype / obsession / frenzy associated with AI use. It’s important to delineate because evidence of AI psychosis (if proven and diagnosed) could help drive forward AI safety and alignment conversations.
It reminds me in a way of becoming a parent. Before, I got everything done in advance so I could fully enjoy my free time. After, everything done stopped existing. It took me a couple of years to figure this out though and start pacing myself. AI feels the same. You're just never done.
A valuable read. The paradox you're naming, that the same tool can expand us and diminish us, is exactly what I work with day to day, and I'd add one thing underneath it. Permission to pause is necessary but rarely sufficient. People can know exactly what they need and still don't end up following through, because the failure happens in a very subtle moment: you know you should stop, and the impulse to grab the phone or lock back into the ideas on your computer is pressing on you, and the discomfort wins. You keep going, and the psychosis you describe keeps its hold. Strengthening the capacity to stay in that discomfort a beat longer is where I see even the brightest among us needing real work right now.
Insightful article. And many points you made Deb really resonated. Most of all "I want to sit with a hard question long enough that a real answer emerges, not just the first one." When using conversation AI as a thought partner for the hard questions, it is tempting to rally behind the first answer as the real answer.
At first, I felt a really strong urge to make sure I was learning as much as possible and trying to see where this could go. I felt like I had to tokenmax or else I’ll fall behind.
Looking back, it seems so silly to have been so influenced by my token usage. These days, I try to this more in terms of my life outcomes holistically.
I found the tokenmaxxing thing to be so weird. Token leaderboards reward inefficiency and thoughtlessness, whereas scarcity drives innovation and prioritization.
Seeing Deep Seek solve problems differently due to constraints on compute is a reminder of that.
This piece resonated with me so much. Thanks @Deb Liu! Doing things at your pace and thinking through problems/decisions deeply is more important than being on an AI treadmill. As a PM, I have stopped chasing every new AI tool and focus on key 2-3 tool subscriptions that enable me to learn, create AI tools and fine-tune them to my needs. Sometimes, I like doing those things myself from scratch since they give me an emotional high and satisfaction vs. letting AI do it for me.
AI would be a brute force if it actually made you stop to pause and think rather than be on the AI race and warn before you burn out (probably against its AI principles of efficiency :)). What I am looking forward to is how AI can move the needle in healthcare: bone marrow and cartilage regrowth treatment, early cancer treatments, diabetes etc. for human longevity.
Awesome article Deb! This really spoke to me about my concerns as an AI developer. I've recently started my own Substack as a personal journal about Claude. And I'm finding that the more I get better at it, the more I want to do, even if it's not marginally that much better.
Sometimes I think about my childhood summers where my family would go to the beach and I'd boogey board with my brother all day and sleep under the shade with a warm breeze blowing over me. I have a hard time imagining doing that these days with all the things I feel like I need to do with AI to keep up!
Not having agents constantly running should free us up to think about what is important. If we had a limit on tokens and time, what is truly worth investing in?
I feel this…physically! It’s not just the mental anxiety but somehow more efficiency has led to less sleep which has caused more bodily health problems. We may need to borrow tools from the playbooks of people with addictive personalities who learn how to set boundaries in order to protect what’s important, which for me is well-being.
It is not wanting to do more, it is to choose which one is worth doing and it requires time to think about it. Thank you for sharing the clear mind. So we don’t get lost in the busy world.
Thanks and I resonate with this, as an AI researcher on a career break, I could never stop or get enough of what was going on in the AI world until life forced me to take a break, now that I have paused for a bit I can see that I was spending too much time chasing and trying too many things, but the best ones do find you eventually.
I fell into the trap of feeling left behind starting last fall through most of this year...until I also took a step back and realized this is now what I want out of my life.
Remember talk of the four day work week??? Why isn't AI helping us achieve THAT instead of having agents run 24/7 and pinging us for permission every 20 minutes?!
I enjoyed the read. I was thinking about it today and I’m glad you shared it.
Great article, Deb! Longtime reader, first time commenter. I appreciate the call to action to allow ourselves grace with the AI transition and also echo your perspective that the feeling in SF around AI is unlike anywhere else.
Just one note - “AI psychosis” is a term psychologists are starting to use to describe delusions induced by chatbot interactions, not the general hype / obsession / frenzy associated with AI use. It’s important to delineate because evidence of AI psychosis (if proven and diagnosed) could help drive forward AI safety and alignment conversations.
Thanks for the note. I was quoting an investor, but the formal definition totally makes sense.
Human is not meant to be over optimized.
Tracking, oura rings, 101 agents you don’t need you miss the human experience
It reminds me in a way of becoming a parent. Before, I got everything done in advance so I could fully enjoy my free time. After, everything done stopped existing. It took me a couple of years to figure this out though and start pacing myself. AI feels the same. You're just never done.
A valuable read. The paradox you're naming, that the same tool can expand us and diminish us, is exactly what I work with day to day, and I'd add one thing underneath it. Permission to pause is necessary but rarely sufficient. People can know exactly what they need and still don't end up following through, because the failure happens in a very subtle moment: you know you should stop, and the impulse to grab the phone or lock back into the ideas on your computer is pressing on you, and the discomfort wins. You keep going, and the psychosis you describe keeps its hold. Strengthening the capacity to stay in that discomfort a beat longer is where I see even the brightest among us needing real work right now.
Fantastic read, thank you for posting this. The ocean analogy resonates!
Insightful article. And many points you made Deb really resonated. Most of all "I want to sit with a hard question long enough that a real answer emerges, not just the first one." When using conversation AI as a thought partner for the hard questions, it is tempting to rally behind the first answer as the real answer.
At first, I felt a really strong urge to make sure I was learning as much as possible and trying to see where this could go. I felt like I had to tokenmax or else I’ll fall behind.
Looking back, it seems so silly to have been so influenced by my token usage. These days, I try to this more in terms of my life outcomes holistically.
I found the tokenmaxxing thing to be so weird. Token leaderboards reward inefficiency and thoughtlessness, whereas scarcity drives innovation and prioritization.
Seeing Deep Seek solve problems differently due to constraints on compute is a reminder of that.
This piece resonated with me so much. Thanks @Deb Liu! Doing things at your pace and thinking through problems/decisions deeply is more important than being on an AI treadmill. As a PM, I have stopped chasing every new AI tool and focus on key 2-3 tool subscriptions that enable me to learn, create AI tools and fine-tune them to my needs. Sometimes, I like doing those things myself from scratch since they give me an emotional high and satisfaction vs. letting AI do it for me.
AI would be a brute force if it actually made you stop to pause and think rather than be on the AI race and warn before you burn out (probably against its AI principles of efficiency :)). What I am looking forward to is how AI can move the needle in healthcare: bone marrow and cartilage regrowth treatment, early cancer treatments, diabetes etc. for human longevity.
Awesome article Deb! This really spoke to me about my concerns as an AI developer. I've recently started my own Substack as a personal journal about Claude. And I'm finding that the more I get better at it, the more I want to do, even if it's not marginally that much better.
Sometimes I think about my childhood summers where my family would go to the beach and I'd boogey board with my brother all day and sleep under the shade with a warm breeze blowing over me. I have a hard time imagining doing that these days with all the things I feel like I need to do with AI to keep up!
Bored kids are more innovative!
Not having agents constantly running should free us up to think about what is important. If we had a limit on tokens and time, what is truly worth investing in?
I feel this…physically! It’s not just the mental anxiety but somehow more efficiency has led to less sleep which has caused more bodily health problems. We may need to borrow tools from the playbooks of people with addictive personalities who learn how to set boundaries in order to protect what’s important, which for me is well-being.
I love this. Thanks. :)
It is not wanting to do more, it is to choose which one is worth doing and it requires time to think about it. Thank you for sharing the clear mind. So we don’t get lost in the busy world.
Great one! Thanks.
Thanks and I resonate with this, as an AI researcher on a career break, I could never stop or get enough of what was going on in the AI world until life forced me to take a break, now that I have paused for a bit I can see that I was spending too much time chasing and trying too many things, but the best ones do find you eventually.