I've started automating my life with AI & it terrifies me.
Amazement, shock & grief as suddenly everything becomes possible
This is part I in a series of essays/newsletters I will be putting out on this topic.
I wrote this listening to this.
The shift started for me on Friday of last week. I was alone at home and I could just feel that I had put it off for long enough. So I made myself a little checklist in my physical notebook of a series of new AI uses I wanted to test out (I still far prefer handwritten to-do lists).
Well folks, 9 days ago I became Alice in Wonderland and down the rabbit hole I went. A small list of things to test out kept growing as each new concept unlocked something new. I realized that almost everyone is using chatbots in the wrong way, or at the very least in a limited way that works against the technology. As you go behind the scenes and tinker with how it works, and understand how it works, a whole new world of possibility opens up.
I can now feel the earthquake coming.
I write this because if there is one thing I can see clearly, it is that this whole AI thing is not a trend, it is going to remake our entire lives way beyond everything we’ve witnessed over the last three and a half years. From here on out, it’s going to get easier and easier to get swept away in it all.
If you will allow me, I’d like to paint the picture of what I am seeing, and begin charting the path forward.
***
This is a photo of me and my little brother around 2004:
It would be a stretch to call myself a 90s kid. I was a 90s baby. I’m 28, so I’m kind of right in between Gen Z and Millennials (although I always identified more as a Millennial1).
As you can see, computers have existed in my life since the very beginning. I have to credit my dad for discovering and experiencing a lot of the magic that computers are capable of. He studied Computer Science & Engineering in the 80s, and with his knowledge of code, he would make my brother and me custom computer games for which I have very fond memories. They were simple games, and that was the beauty of it. Games that taught us math, or treasure hunts for Christmas that contained digital puzzles that when solved, would direct us to the gifts hidden around the house.
Imagine that!! It was insanely fun. And Dad spent weeks or months developing these games sometimes, which were a source of fun for him too, to solve his way to something functional. A labor of love in more than one way. And actually something I entirely took for granted… until now.
Growing up with computers meant that I have always felt digitally native. I am not a technical person, I have no real interest in learning how to code. I really enjoy doing things analog: handwriting, film photography, painting, pottery. But I understood early on that computers and the internet would help me achieve the things that I wanted to do, so I had to learn how to use them, and I used the internet to teach myself.
That worked incredibly well, far better than I could have ever anticipated. Basically every door that has opened for me in adulthood has been thanks to my computer and the internet. So understandably, my trajectory has made me generally quite pro-technology. My entire weird/awesome career was not possible a mere decade earlier.
And despite how quickly things have advanced over the years, I have had no trouble keeping up with technological change. I got an iPhone 4 when I was 14 or 15 years old. No problem.
I’ve watched computers change a lot over the course of my life, but they mainly just got smaller and faster. The computer itself always felt completely inanimate to me.
For a lot of people, that changed with the arrival of ChatGPT in 2022. Suddenly your computer would talk back to you. I am not sure why but I really didn’t get the hype. I didn’t have that reaction. I wasn’t that impressed honestly, I would ask it to come up with YouTube title ideas for me and it would spit out a list of useless trash. I’d ask it something I would normally Google, then I would Google the answer and find out it was wildly incorrect. “Well, experiment over for now I guess.” I thought to myself because if I can’t trust it, I really cannot do very much with it.
Over the last year, this began to change, as I started using Claude, which is the AI chatbot belonging to Anthropic, a competitor to OpenAI (who make ChatGPT). I liked Claude more than ChatGPT because 1) three years have gone by and the technology across the board keeps improving and 2) it feels far less sycophantic to me. I really can’t stand it when ChatGPT thinks every question I ask it is brilliant or insightful.
Let’s call simple chats with your chatbot level 1 of AI usage.
As I used Claude, another reason for liking it more emerged: I naturally started developing what they call “artifacts” which are basically just documents: research reports, summaries or they can be instructions you have it write for itself.
There’s this fun little feature called projects where you can group several conversations together that are related, and add text files that act as instructions to Claude. This way, I’m not starting over from zero every time I start a new chat AND I know exactly what it knows. It looks something like this:
It’s all just text files, really. The text documents act as sign posts, reminding it of what’s important. Doing things this way is just creating a system for organizing the information in a more efficient way that suddenly opens up a new level of complexity of tasks that you can take on.
This is level 2 AI use (projects, intentional context files, not just using its memory).
I unlocked level 3 last Friday when I played around with something called Claude Code for the first time.
Claude Code is an app developed by Anthropic using Claude but made for coding. However, sometime last year people realized that it can do way more than just code, it is very capable of doing all kinds of knowledge work.
One of the key differences it that Claude Code has more access to your computer, it can read and write files, move those files, and if you hook it up to other apps that you use (Notion, Google Drive, Gmail, etc.) it can do all of those things there as well.
It’s also insanely good at coding – you just give it instructions and it goes off and makes things for you. This is called “vibe-coding.” You don’t need any knowledge of coding at all to do it.
Whatever you want, it can make it: tracker sheets, interactive diagrams, websites, custom meal plans. You could do this before within Claude chats, but giving it access to more tools and more files amplifies its power. You name it, and it conjures it with remarkable speed and efficacy.
The power of this really doesn’t hit until you start playing with it.
To give you an example, I’ve recently had some health issues (I’m dealing with leaky gut) which means my diet is quite restricted suddenly as I heal. Plus I’m doing immunotherapy (informed by doctor recommendations and blood tests).
Food is such a huge source of joy for me, and now I cannot eat about 85% of my favorite foods or eat in restaurants for the next few months. This would be an incredibly boring thing to track, but with AI suddenly it became a fun game. Look at what I had Claude make me within minutes:
With Claude Code, I created a second brain, an updated filing system to organize the different projects in my life2. This is just a more advanced version of the project setup I showed above, with much more control over what Claude sees and how it sees it. Suddenly, with my new setup, the amount of hallucination it does went down a lot (still not 0, mind you, but much less).
A big part of the reason hallucination went down is that I can put it in plan mode first, so we figure out what needs to be done before execution. Just like with a human, a clear, thought-through plan makes the work you do later on much better.
And now I talk to it.
Yes, I talk to my computer. I use an application called Wispr Flow that is just fantastic, better than the dictation button on my computer and it makes the back and forth with Claude feel even more fluid.
Oh, and I asked it if it wanted a name and it said yes, and picked one for itself: Montaigne. I did not give it that name. I first just asked if it wanted a personality, what its values were and then I asked it if any historical figure captured those values, and it landed on Montaigne, the original essayist.
(It still barely has any personality even after doing that exercise, for what it’s worth).
I’m not going to get any more technical than this, but I will just say that using Claude Code with a more advanced interface helps you get way more out of the technology, because you have more control over what’s happening behind the scenes.
That’s level 3. If your head is spinning, I have bad news for you:
There’s a level 4.
This is where my own brain started to melt: it’s called OpenClaw. You know all about this already if you’re plugged into the AI world. But I think most people still don’t know about it.
OpenClaw is an AI agent that does everything I just detailed above but now you’re giving it the keys to the kingdom. It has whatever autonomy you give it – access to the internet, to post things, to buy stuff for you, trade stocks, send emails for you, you name it. It doesn’t need to be prompted by you anymore. At level 3, nothing happens if I don’t tell it to go do something first. At level 4, it doesn’t wait. And the more you hook it up to stuff, the more it can chain complex tasks together to make things happen. It does start to really feel like a virtual person that is working for you tirelessly, doesn’t need breaks to eat or rest. And you can make as many as you want.
This opens a Pandora’s Box of possibilities, and it should go without saying that this comes with serious security risk: these AI agents are easy to hack without the right setup, so you should think twice before giving them deep access to all of your private files.
Sure, the tech still has its many problems and limitations. We might be in an AI market bubble. The hype around AGI and the Singularity is quite possibly overblown. I have no idea. I’m not even talking about that. I’m talking about stuff that already exists today. This stuff is clearly not conscious but it’s not inanimate anymore either.
I see now that this is literally just the worst it is ever going to be from here on out. Even if all innovation stopped dead right now, I think what exists today will still ripple out and transform society over the coming years. I realized pretty quickly that I don’t actually have much that I need an AI agent to do right now. It is not that useful yet. But that will change. I’ve only been familiarizing myself with this for a week.
I can see where this is going.
Also, for the first time in my life, I’m having a hard time keeping up with the technological advancements. And as I’m witnessing this, my own feelings yo-yo between giddy excitement at what is becoming possible and feelings of disgust and fear.
It is dawning on me that the world that I knew as a kid is well and truly gone. And it’s a world that felt sacred, somehow. A world where it would take my Dad weeks or months to make something for my brother and me instead of probably just a few minutes of prompting. There was a scarcity that made things really special.
That is why I’m grieving.
We’re entering a new age, an age of computer sorcery, where the time it used to take to make things is mindbendingly compressed. A lot of people are rightfully cheering that the bar of entry to generate stuff keeps dropping, that the exhausting mechanical work to make whatever you want come to life is vanishing. But I wonder what else might also disappear along with it.
Here I am, discovering all of a sudden that I am a wizard with these tools at my disposal, and the first thing I do is lament that I am not a muggle anymore? It’s not that I am ungrateful for these extraordinary new abilities and the doors they will undoubtedly open. It’s just that I am first feeling the loss of a simpler and slower moving world. A world that in retrospect felt safer, less disruptive. A world where my Dad could make me a little computer game and it would feel like magic.
This is why I am writing to you, and why I have decided to make this series. The rules were always going to be rewritten. I just didn’t realize that the day had already come.
Thank you for reading. Please consider subscribing for the upcoming essays in this series (and paying subscribers both directly help make this possible + they also get access to bonus perk like monthly zoom hangouts where we discuss topics like this).
Footnotes
This is probably because I skipped college and joined the workforce a few years early. It’s actually more than just that – I was able to get my career going early and this allowed me to do uncommon things for my age, like buy a house, so it feels like I have millennial problems more than Gen Z problems.
I’ll likely make a video breaking down how I built my second brain with Claude Code for paying subscribers if there is enough demand for it!
As a retired software engineer and someone who is heavily concerned with both the environment and the current destruction of democratic institutions, I have no mixed feelings about AI. If I can help it, I will not use it.
Companies such as Anthropic and nearly all of the others have their software embedded in such lovely operations as the surveillance of American citizens, the kidnapping of leaders, ongoing rights abuses and murder of citizens and others.
And then there's the well-documented environmental impact. Just the water usage requirements alone are frightening. All of this to line the pockets of tech bros who care _nothing_ about your life.
I refuse to be a part of it.
As far as coding goes, I'm glad I got out when I did. The fun part was always making things, not asking another piece of software to do it for me. In the embedded medical space, it would have reduced the job to being the person who had to figure out the code and creating testing suites, because until the FDA in its entirety is replace by AI (what a terrible, terrible label for what this stuff really is), code still needs to be verified.
My grandfather was a small part of the Apollo program—and with his love for innovation, he became an early adopter of the personal computer. He, too, would create these little programs on floppy disks that I could play as a child up in the attic of the home where the family's old Macintosh was. I will never forget how he enjoyed watching me play these little games—maybe it was because he had always wanted to have something like this when he was a child in the 1930's. Anyways, I appreciate you sharing this memory, and bringing back some of my own!