42) AI, Self, & the Discourse

Behind-the-scenes building Vambrace AI, a company on a mission to figure out its mission. Please pardon the stream-of-consciousness style. Subscribe to follow along or visit the site here:

(typos are to verify my humanity)

Introductory Remarks

In last week’s post, Distribution, I reflected on the importance of distribution—especially in a world of product-abundance. How you find, talk to, and interact with your audience is critical to any business success; and the biggest challenge in that whole operation is making them actually care. Nobody cares as much about what you’re doing as you do, and that can be a difficult pill to swallow, particularly given the emotional investment typically associated with entrepreneurial endeavors. But that’s where resilience and grit come into play. Moving on!

(this is a long one)

AI & Self

This post was inspired by Rex Woodbury’s post, Why Does Everybody Hate AI?, filtered through the prism of my own experience. I intend to recount my own lived experience with AI, and then explore some of the current tensions I feel culturally and internally given my extreme use of the technology.

Let’s go chronologically and see if anything relevant comes through.

November 2022: The ChatGPT Moment

ChatGPT 3.0’s release on November 30, 2022 is the consensus breakthrough moment for AI. This release led to somewhat widespread adoption, even more widespread acknowledgment of the technology, and compelled ChatGPT to be one of the fastest growing consumer mobile applications in history.

My personal reaction was mixed. It was obvious that this was pretty powerful and neat technology, but at this point in time it was still largely science project-ish, and it definitely wasn’t ready for primetime use. It also wasn’t yet connected to the internet, which was a clear deficiency in its ability to surface relevant and current information.

But it was still fun to use and I enjoyed some low-stakes applications around it helping me build workouts and thinking through diets and stuff. I think I had my mom write it a draft company-wide email for her engineering firm. The tool was amusing and felt truly novel, but still more of a curiosity than a mechanism of industrial change.

September 2023: Internet Connectivity

On September 27, 2023, OpenAI announced that ChatGPT could now browse the internet. I don’t personally remember this specific moment, but it was obvious that this was a big breakthrough in terms of its utility for and relevance in daily use. I might have started using the tool more for alternative search at this point in time, and I think around this time was also when google was really spooked and felt behind and stuff. General reaction was still: this is neat technology that I can see myself casually using, but I still can’t trust it in commercial settings.

May 2024: ChatGPT 4o

On May 13, 2024, ChatGPT 4omni (4o) came out, and I remember that as a pretty significant stepwise change in performance. With 4o, I felt like I was getting fairly high-fidelity feedback and maybe late high school / early college-level responses about complex topics. This was the first time I think I started using ChatGPT to enrich my own personal workflows within venture capital—albeit still in a very lightweight manner.

Despite its newfound intelligence, I remember hallucinations still being a massive problem here—and there were several specific moments where AI would confidently cite statistics from some source, but then if you went and traced that source you weren’t able to actually find the statistic. That was a scary experience, and I think a big part of the reason why Perplexity so quickly came to the fore and was successful in carving out a niche within more research-oriented AI applications.

December 2024: Projects

This is my personal timeline for Projects, which I believe were also launched in May 2024. But I vividly remember a conversation I had with a friend who told me about Anthropic’s projects feature (I remember not loving or not being aware of ChatGPT’s projects feature at that point). Projects basically allowed you to upload your own source material and then AI could generate responses rooted in your material. It still wasn’t perfect, but this was probably the first real mind-blown experience I personally had with AI. I remember using it to help me with work that would have taken many hours, and now I was knocking it out in minutes.

I think this also made me feel a little bit of obsolescence and sadness, because my entire professional life up to that point had been about identification, collection, and synthesis of information across many different data sources—and now AI was pretty much doing that automatically. But, at the same time, it wasn’t like I was super passionate about clicking around on different tabs in my browser and then writing notes and paraphrasing and stuff. The productivity gain more than justified the professional discomfort—and I started to think about the technology in terms of its continued developmental trajectory. It was also just so inexpensive—shockingly inexpensive. What was the catch?

Key learning: context is everything.

March - April 2025: Cursor

AI became something we were interested in leveraging internally at the venture firm I worked for, and so I started experimenting with high-level “coding”-based automations. Specifically, I was trying to extract company names from webpages and then auto-send an email out to our team. Ironically, I never actually really built a high-quality and production-level system for this task. But I was introduced to Cursor throughout this process. Cursor blew my mind.

I am generally tech-aware and thought of myself as technical-adjacent. My degree is in statistics and most of my work was done using R (some say it’s real coding, some disagree), and I had taken some introductory coding classes, but I wasn’t proficient. There was a world, though, where I would have loved to learn how to actually code—and I think statistics taught me how to think structurally about data and systems.

Cursor let me walk through, in excruciatingly elementary detail, how to setup a project, how to build an AI funnel, how to launch a website, etc. It was still super immature, and I think I’m only now realizing how limited it was—but there were still these a-ha moments where I felt like I was blind my whole life and now I could see (literally). I still remember the first time I just got it to say, “This is Project Pearl” (or something like that), made me shout for joy (again, literally).

More importantly, I think this was the first time where I felt like I could really build things. There’s this discussion in tech and startups about being technical vs non-technical—and I had always been a little insecure about being “non-technical,” despite my interest in and appreciation for technology. So Cursor gave me a taste of a technical existence.

Even more importantly, Cursor made me feel equipped with the tools and skillset to actually build real things. And then the internal calculus for me was, well, would I rather sit around and watch people build really cool things, or should I just dive in head first? This is when I first started the user feedback thing (bad idea, but an incredibly useful learning experience).

Key learning: the lines are blurring between technical and non-technical.

September 2025: Claude Code

I had heard about Claude Code for quite some time, and then finally after a startup bootcamp in NYC in September (s/o Precursor), I felt compelled to take the leap. The terminal was scary, but eventually I grew to love it—and now I can’t live without it. Claude Code was another stepwise jump in how I used the technology. It also created a whole new world of possibility around contextualization since it was basically AI that had direct access to my machine. So I could feed it files on my computer, and then it could take action based on my information.

Key learning: every single bit is context.

January 2026: Clawdbot

I wrote about this recently, but Clawdbot also took the world by storm in January. It was the first reliable-feeling personal assistant. The two main breakthroughs were (1) the ability to interact with it remotely via iMessage or telegram, and (2) a somewhat-persistent sense of memory.

Key learning: persistent memory is the biggest challenge in the industry right now (where memory is defined as session-agnostic context).

March 2026: MCPs & AI Command Center

More recently, I’ve been expanding into more technically-efficient personal productivity enhancements to help equip Claude Code and Codex with context about the company. Specifically, I have a one-click tool call called /sync that reads in all my granola transcripts, my recent git commits, vercel pushes, and then maps that to linear issues—and AI does it all. I’ve also started using AI to handle my Instantly outbound campaigns, which has been a massive unlock as well.

Key learning: point-and-click software interfaces are going to be a thing of the past.

AI & the Discourse

So, why go through the pain of explaining all of this? The main point is that it’s been a wild journey, full of twists and turns, but that has consistently been up and to the right in terms of my personal productivity and my amazement of the technology. I’m probably 20x+ more efficient than I was 4 years ago, and I can do things that I never thought I’d be able to do. So, I’m in awe of this incredible technology, so much so that I quit my job to pursue this full time. I’m as all-in as all-in can be.

BUT, people absolutely hate AI—and I feel that in my personal life. This isn’t a woe-is-me point, but I am bashful and sort of ashamed when I tell peers that I work in AI consulting. It feels like a dirty industry. It feels like I work in adult entertainment or gambling or alcohol or cigarettes or some vice-based industry like that. And I don’t really know why? Well, I do know why, and that’s the problem.

AI & Work

(These beliefs are developing and shifting as time unfolds)

Obviously, there are very real threats of AI-induced job loss. I understand that that is really scary to many people. As a techno-optimist and a slight anarcho-capitalist (in industries where regulation isn’t explicitly beneficial to end-users), I don’t think anybody should be doing anything that a robot can more cheaply and more effectively do. That just makes no sense to me.

It’s important to note that I believe this for non-artistic and/or culture-based product categories. There will always be intangible value in handcrafted leather goods (s/o Hermes), but there’s no real hand-crafted equivalent for transportation or FP&A or etc. Maybe the real distinction here is binary-outcome vs non-binary-outcome jobs? Or value-additive vs procedural jobs? I don’t know.

But, the main point I want to make here is that, I understand how scary AI can feel when it potentially entirely upends the societal hierarchy and implicit social contract of the late 1900s / early 2000s American Dream. That’s scary, I get it—I feel it! I oscillate between extreme optimism in how much work is out there for me to do in the coming years and extreme pessimism and apathy around feeling like AI will just do anything for anyone perfect at the time of consumption—and so like then what’s my purpose?

AI & Art

As the son of a musician, I also understand the tension around AI’s impact on artistic and creative output. It’s never been easier to create high-fidelity and decent-sounding AI songs. It’s also never been easier to create some artistic photo and video outputs. For me, it’s been super fun because I could never make songs before, and now I can make full-length albums. To me, it’s just another modality for my personal creative expression—and I don’t really think that the songs I make can ever really compare to the music my dad makes.

In this category, I think it will be similar to what I said above, whereby low- to mid-tier musicians and artists will probably be wiped out, but high-tier craftspeople will retain authority and ownership of an engaged audience. There is something very human about consumption of creative artistic works, and I think we often forget that consumers are complicit in the creative contract.

Did consumers boycott streaming because it was so destructive to so many musicians? No, I don’t think so. And so to a large extent, it will be up to consumers to continue supporting musicians they love—and I do think that there is value in understanding the human behind music and art—and I don’t think abundant AI-generated content will entirely displace that.

This is actually where I think Infinite Jest is sort of relevant, because it’s main contention is that, could people really abstain from watching a piece of entertainment that is compelling it literally captures them until expiration, if they knew it existed? Is that so dissimilar from an infinitely scrolling TikTok feed? Are we all already just wasting our lives away, captured by enormously biologically appealing slop?

Last thing here, I also don’t think real artists are purely economically-driven (or even economically-driven at all), and so really AI probably shouldn’t have any bearing on their output. If anything, maybe now they can make low-cost music videos or compress the time from song creation to mastered production to share with people. I don’t think that’s necessarily a bad thing?

AI & Education

Another challenging pillar of the discourse is AI’s relationship with education. I’m torn on this one. I recently bought 3 textbooks on proofs and algorithms and writing, because I do think that, as AI becomes more effective, I want to make sure that I’m preserving my mental and cognitive agility. But at the same, I have this intelligent 24/7 personal assistant that can help learn about literally anything. So how do I reconcile the power of this technology in learning—I mean it’s already helped me learn so much about technical development—with the uncertainty of its long-term impact on my cognitive health?

I also do agree with the discourse that the social contract for college education is starting to slip. As I’ve said in the previous sections, I do think that the bottom half (or so) of colleges will probably be wiped out in some form or fashion. I also think that we’ll see new AI-first universities pop up that take a completely different approach to education in an AI-first world. I’m not sure what those will really look like. But I do think that our education structure is due for a reckoning and restructuring—and I understand that that is also really scary and unfair to many many people.

AI, Self, & the Discourse

Pain

So, I guess, what I’m trying (ineffectively) to reconcile here is (a) the immense pain and suffering that I’m sure AI will indirectly and directly inflict upon millions (if not billions) of people, with (b) my intense excitement for it and optimism around its use—and then also my direct participation in its adoption. Am I horrible person? Or is it better to have a well-intentioned person driving one of these AI cars? But what does it even matter if I’m well-intentioned if I can’t actually bring about those positive intentions?

So I guess I struggle daily with feeling so excited and joyful about doing work I love with people I respect, in categories that I thought I’d never be able to enter. And then I feel sort of immense underlying and peripheral waves of guilt about my complicity in future pain. But then, in a competitive market it is theoretically zero-sum, and so why not me?

Responsibility

Am I complicit in the plight of millions, if not billions of people in this world? Or am I well-intentioned techno-optimist just trying to do his part in making the world a more efficient place? Is it bad that I am so excited and energized each day to use this incredible technology to help myself and companies become more efficient? Is status quo the alternative? Why can’t we realize that there are probably jobs on the other side of this that we can’t even conceive of right now?

Success

Will we enter a post-economic / post-capitalist society? Will money lose its meaning? What will it mean to have an abundance of low-cost intelligence? And then what does it mean to pursue the American Dream against the backdrop of such broad economic and social uncertainty?

Every single day I’m torn by the paradoxical feeling of living in the most disruptive time in human history with one of the most transformative technologies of human history, and feeling like I’m not doing enough to capitalize on it economically—and then feeling like I’m an idiot for not figuring out how to capitalize on it economically—and then feeling apathetic about the eventual lack of opportunities to capitalize on anything economically—and then wondering about the implications of not being “successful” on my romantic and life prospects.

But then I sort of snap out of this thinking and remind myself to focus on the day ahead of me, and to just do what I can today. And that there are millions of companies that will take time to adopt this technology, and there will be new opportunities and modalities of consumption and value and exchange that we can’t know of right now.

Culture

I like reading books written by humans, and listening to music made by humans, and watching things about humans (s/o Love Island and Traitors). AI can’t take that away from me, right?

Physical Health

In a world of digital abundance, really the main “status symbol” will be physical health. I think this kind of explains the recent movements towards superficiality and “looksmaxxing” and intense physically-oriented health protocols and ozempic and stuff. In a world devoid of economic human value, there is still aesthetic human value—and AI can’t stop someone from looking at you and thinking you are beautiful and handsome.

But “looks” and beauty aside, I do think physical health will become much more important in the coming years—and I also think we may have more time, ceteris paribus, to engage in physical activities that help us maintain and improve our health. Physical outlets also allow us to socialize in the real world with other humans, and that is going to be more important than ever.

Relationships

And then beyond physical health, I think the connections that we form with other people will become more important than ever. You can’t AI away looking somebody in the eyes and really sharing a moment with another human. So I think there will a premium on in-person interaction and the social connections that we form, and (hopefully) the romantic connections that we form. Who would you want by your side as the end of the world looms? (s/o Bell X1, The End Is Nigh).

In a world where the industrial operations of society, writ large, become largely automated, that frees us up to focus on uniquely human things: loving, living, crying, laughing, leaping for joy, striving, supporting, and doing things with each other. I think empathy and social intelligence will become one of the most valuable currencies of the future.

Looking Forward

Things got away from me a bit today, but I’ve been meaning to say a lot of this. These feelings have been percolating around my mind for the past few weeks and months, and it’s been really difficult to figure out how to move forward with kindness, optimism, and enthusiasm in an industry that is so hated and so transformative.

Oh well. Have a great week!

Sincerely,

Luke