It’s Not All Slop
TL;DR: AI makes problems easier to solve with technical solutions. There are lots of unsolved problems in the world in need of a technical solution. That’s good, let’s roll up our sleeves!
There’s a gap between how AI feels to an engineer and to someone who doesn’t write code. In 2023 when I first was using ChatGPT to generate code for me, I told a friend that it felt like I was working in the future, like I had a Mass Effect omni tool on my arm while I was programming. At work and online, I hear a similar perspective. AI is making us faster, allowing us to solve harder problems and get more things done.
i know everyone's shilling opus 4.5 left and right but honestly coding has never been this fun. i've been doing iOS development for over seven years now and the speed at which i can go from idea to working prototype is something i couldn't have imagined even in my wildest dreams
— “paula” (@paularambles) January 10, 2026
Meanwhile, the opinions I hear from outside of tech aren’t as excited about AI. In one conversation I was asked “has AI actually changed anything about your work?” and I replied “yes, almost everything,” and I don’t think that person believed me. This doesn’t surprise me, because I’ve had two distinct AI experiences - the one where it makes me a super fast and much more effective coder, and the one where everything on the internet is slop that sounds the same with samey-looking pictures. I imagine if your experience with AI is tilted towards the slop end of the scale, and if your profession is not one where AI is an obvious booster, then AI is of course going to feel like a useless toy that makes electricity bills go up and drains the Earth of all its water.
There’s another gap between tech and non-tech I’ve observed for a while, since before the LLM era. This is the gap between problems I know that good tech could solve and the incomplete current solutions to those problems. I’m optimistic about AI because I see that there are so many low-hanging fruit problems left to solve, and now AI has made a lot more people capable of solving them quickly.
Let me talk about what I mean when I talk about this low-hanging fruit. Way back when I was a baby data scientist (or as they liked to call it, Advanced Analytics Developer) at a Fortune-500 company that did something like 70% of its billions of dollars in sales by phone, I came across this blog post from 2016 by Mister Money Mustache with a story that seemed to capture how weird these low-hanging fruit problems look to coders. The author tries to order concrete for a construction project and finds that it can only be done over the phone, and only the day before, and only after 12pm.
Concrete is not a niche cottage industry like homemade salsa – this is a $35 billion chunk of the economy that is critical to building almost everything. A single loaded truck carries $1500 of the stuff, and there are 50,000 of these trucks in circulation in the US. And yet not only have they not discovered computers, even the concept of a notebook with two separate pages (“today’s orders” and “tomorrow’s orders”) was foreign to this outfit. This story is just an extreme example of a market opportunity that is still fresh and ripe in our society as a whole. We have computers, but a deeper understanding of how technology works is still rare. Almost every big company that I’ve observed is still clunking along, trying to adapt to technology rather than fully benefiting from it.
“A deeper understanding of how technology works is still rare” is why local governments, small businesses, even large corporations can’t hire people with the technical ability to build and maintain even the easy stuff, and that means a lot of problems that should be easy to solve are just left to sit. I’ve reflected on the absurd gap between what I know tech can do and what is available when I go to my local DMV to register my car. “Do you have proof that you paid us taxes?” they ask me, and I imagine saying “do you not have relational databases back there? Why do I have to prove to the city that I paid the city? If you just let me go back there to mess around with SQL for a little while I’m pretty sure I can figure this out…”
I’ve also had some experience, either personally or through close friends and family members, navigating the systems for applying for benefits or aid for veterans, the disabled, and refugees. I often find the applications to be complex and hard to understand, and I’ve been lucky in my access to education, resources, and support. I suspect that there are people who tangibly suffer because of these inefficient processes making it difficult for them to get help. However, I have seen an example of how someone made a technical solution for this kind of problem, so I know it’s solvable! Nicolas Bouilane noted that the form you have to fill out to change your address in Berlin “sucks. It sucked 8 years ago and it sucks now. It’s ambiguous and confusing and I hate it, so I made a better one.” Awesome. I’ve never lived in Berlin but the frustration in his words about filling out this form is exactly the kind of problem I’m talking about. This guy made a tool that makes clearing a bureaucratic hurdle easy and accessible and he put it online for people to use. Ever since I read this post, I’ve wanted to do something similar. As far as I know he didn’t use AI to do it, but now with AI, the bar for people like me to do the same kind of thing just got a lot lower. I think that’s great news. Let’s use our newfound AI powers to do more of this.
This all leaves me feeling pretty good about what AI can do for the world. I see a lot of problems that need solving, and I see a tool that lets me solve more problems with tech. When I taught an intro-level Data Science course, I told my students often that coding is a superpower. I’ve lamented that so many of us in tech use that superpower on stuff like selling ads (or much worse). Now with AI, I have more time and more ability to build solutions. My personal call to action is when I look at my vibe-coding backlog, I’d like to prioritize the things that can help people, even if it’s just a few people in my life (because the bar is so much lower, let me automate that for you!), and when I encounter an arduous process in the wild (why is there no complete documentation for how to register an out-of-state leased vehicle in my home state, argh) I want to think if I have the power to make it easier for someone else.
I’m hopeful that over time, AI lowering the bar to solving problems with tech means that more problems get solved, and more people get to feel the amazing futuristic parts of the AI era that I’ve been enjoying while writing code.