AI has been stealing my joy at work.
I chose to become an investor to build relationships with incredibly talented people. To regularly be shocked by a unique point of view and ambitious plan to change the world. To follow my curiosity and see what versions of the future I believed in (or not).
It occurred to me the other day: when I use AI too much, I enjoy my work so much less. I miss the moments I’d look up and realize hours had passed while falling down rabbit holes of markets, new tech. New AI norms in venture (promising greater speed, deeper analysis), have compressed all of this into reviewing a stack of research memos created by a soulless all-knowing machine. All of my conversations and research get flattened into this growing to-do list of sameness. Totally devoid of the signal I used to get from following my energy - where I wanted to learn more, the random connections to other areas of interest.
Maybe I’m more efficient, but I can’t help but feel AI has lobotomized the part of the craft that actually led me to better decisions.
I know this isn’t unique to venture, so I wanted to take a closer look. Can we preserve the joy we derived from our jobs before AI? I think a big part of it is putting the technology in its rightful place - which, to me, falls into two dimensions: public leverage and private offloading.
Public Leverage
AI can do work we can’t easily comprehend - solve mathematical proofs, discover valuable new molecules. We need this and need to keep making it better.
OpenAI recently shared that internal versions of their latest frontier model solved 10 major open problems in math and science. It was able to verify solutions computationally, which is what matters for major advances.
I know AI extends our capacity to explore and create. There’s a joy in seeing what we’re capable of, in wielding powerful tools that can raise our agency and impact with intelligence beyond our own. While it’s raising the ceiling in these ways, it’s also automating away the repetitive admin and low-agency work we never wanted to do anyway. Both represent real leverage.
However, as we build our personal AI stack, we need to spend more time in the middle asking and answering two questions: Where do we want to invest our attention? What do we want to prioritize? The middle is where our unique perspective lives. When AI amplifies this, it can not only make our work better, but way, way more fulfilling. Getting it wrong is a disaster.
Private Offloading
This is when AI argues a point, conveys conviction, or understands something on your behalf. This is the dangerous addiction with terrible long term consequences.
YC applications have grown 60% in length (word count) over the past three years. Founders are answering diligence questions with AI. Board members are sending AI generated follow ups after meetings. Hiring managers are struggling to evaluate candidates’ AI responses after running AI outreach campaigns to surface the candidates in the first place.
In these cases, the content (spoken or written) carries the authority of its author and is supposed to index the author’s quality of effort, thought, and conviction. We need verification systems not to answer “is this correct” but instead “does the human behind this work understand and believe it themselves?”
Everyone is using AI to just hit the ball back over the net because human brains are designed to avoid effort. As a result, we’re losing the sense of accomplishment from a job really thoroughly done. The critical thinking that ultimately gets us to the biggest win is sacrificed in our acceleration of activity with AI. The joy of truly connecting with a person or an idea and feeling the change in yourself over time is lost in the black mirror of today’s single player LLM engagement.
If we don’t find societal norms around public leverage vs. private offloading, we’ll build a world where we don’t know if the humans around us are capable or knowledgeable. In efforts to maintain pace of AI adoption, I worry even the most accomplished will fall into patterns of offloading that make all work disconnected drudgery.
Ironically, the more accomplished you are, the more you fall prey to grindslop.
There is a correlation between private offloading and autonomy at work. The higher you are on the professional ladder, the more your work is easy to offload and hard to formally verify. This is a paradox because the people with the most professional autonomy are also most likely to have crafted a job that brings them joy derived from a sense of agency and accomplishment. But, management and constant synthesis feel painful. Isn’t it nice to have a crutch? AI is seductive in how it promises to dull the pain and accelerate your world. But it’s clear now that the more you rely on it, the more you erode your intrinsic motivation to do the work. And, the most talented folks are starting to ask, “What’s the fucking point?”
Private offloading is more pervasive and much more toxic. We’re all reading so much writing these days that was created by and representative of no one. We can’t use it to assess how someone has wrestled with an idea, how firmly they stand behind it, whether we should hire them, invest in them, or support them on the basis of what they say.
The diligence questions being passed over to AI were intended to help me learn how a founder understood the risks, their customer, what it was going to take to build their thing. Once this content is separated from its human, it’s no longer evidence of belief or insight. It was no fun to create, and it’s certainly no fun to read.
Writing is thinking. The act of authoring is what makes you accountable to the details of a proposal. That accountability is what makes the act of building or doing any of this meaningful. It’s what strengthens your thought process, allows you to encounter challenges before they arise in real life, consider all possible pathways. Decide to not do something at all. Or, really decide to do something with full conviction.
This piece started with a suspicion about how I was feeling. Writing it will change the way I work. I no longer want to read through material asking whether it sounds right to me. I will ask whether the human behind it actually understands and believes in it. I will shift my focus back to the process that created the writing or the research in front of me. Not the artifact as a product to somehow verify.
If you’re a founder building software, AI will help you plan and architect solutions you cannot yet imagine. But don’t shortchange yourself on the joy that comes from your vision for the product, and the responsibility you take for its impact. Especially if you’re collaborating with your team to define a roadmap, your customers to really get a grasp of their pains and needs. The reason to even embark on building a company - one of the hardest things you can do - is to find your conviction on how the world should work, and then make it so. The output is worthless and miserable to create if it’s not an index of your conviction.
I have my own version in venture, but I think it’s true around the industry. As AI continues to be more and more capable, we need to be in constant search for the right leverage to compete without losing the idiosyncratic edge that could take us to the next level. My hope is to know the difference and preserve joy along the way.