So over the past year I have been using or asked to more frequently use AI in my day to day work. This has meant it’s been used for any of the following: being a rubber duck, my own personal junior dev, task organizer, researcher, tester, or on a couple occasions vibe coder.

These have all had various levels of success and failure. While I would love to go into how I could make improvements across the board, I wanted to instead go over some thoughts about sane, rational uses that I think AI fills in nicely. I don’t think any of these should be spicy takes. Just more practical and people centered.

Anyway:

1. It shouldn’t replace people. Or be used as a justification to eliminate positions
2. AI slop shows why people and creativity are still needed.
3. AI can help everyone improve their workflows.
4. AI can be useful for neuro-spicy people. (This is its own whole topic for another post.)

It shouldn’t replace people

This is probably the spiciest take because it does bring into question capitalism’s function if no one has to work. I don’t have answers, I’m not here to argue the pros and cons of it either. Merely to point out that we might be putting the AI horse before the cart. While it does a lot, AI still needs an inordinate amount of babysitting (more on that for number 2). We get told an awful lot that AI will mean we don’t have to work, or mean we only need to work a sliver of the amount of time we work now.

Cool.

On the surface that sounds amazing. I would have all sorts of time for my family, my side projects, my hobbies! What no one says though, is that if no one needs to work: Where am I getting the money to pay for my groceries? Pay my rent/mortgage? Fuel my car?

I’ve yet to hear a satisfactory answer that isn’t some kind of handwaving or deus-ex-machina that it’ll just sort itself out and we’ll all be fine.

Like a lot of ‘move fast and break things’ technologies, until it stops to think and address the consequences, people are going to be necessary. To babysit AI. To earn money to then spend back into the system.

AI Slop Demonstrates Why Human Creativity Cannot Be Replaced

Ugh that’s a long subhead. Does what it says on the tin though, and is probably the most spicy take here.

So much of what makes AI seem ‘real’, is that’s it’s been trained on everything available on the internet. That’s a lot of stuff to work from, but it does not have the ability to develop anything new or create it’s own style. It can only ever imitate from things that already exist.

It cannot replace or out-do actual creativity because there is no rote way to be creative. What works so well for AI in development only hurts it in art. It cannot extrapolate something from a style and adapt it into a style all it’s own. There isn’t solid documentation about how creativity works like there is with code.

At the moment, that’s something that is still a uniquely human quality.

AI Can Improve People’s Workflows

I know there’s some evidence to suggest that AI doesn’t make us all that much more productive in our work. That’s fine. Productivity in an of itself isn’t a great metric if you also want to have a functioning soul. Sure AI can pump out a vibe coded app in a very short amount of time, but there’s likely going to be a lot of time spent afterwards reviewing, fixing, and filling in gaps in that app that the AI missed, forgot(?), or just plain messed up believing it was doing the right thing.

What if instead it isn’t just about productivity, but your effectiveness? AI can be a very useful assistant, or processor, or pair/rubber duck.

Tools like agent.os can help make AI agents a lot more predictable. Yes, it adds a number of markdown files, but it gives the AI specific and human readable/editable guide rails. That predictability does a lot to make me feel like the work is more effective and stable than vide coding could have at the beginning of 2025.

AI Can Be Useful for the Neuro-Spicy

I would preface this section to say that I don’t believe that AI is alive in any way or that anyone should believe everything it says.

That said, I have been using AI agents as a means to beat some of the ‘fun’ executive dysfunctions of ADHD. My brain tends to get hung up on a number of things: starting tasks, finishing tasks, focusing in on work, hyperfocusing on the wrong or unrelated task, or being paralyzed by overthinking. AI does not have that problem.

It’s lead me to make processes that help me work around the concrete hurdles of ADHD. Helping me kick off work, or creating a defined list of items to do around a particular task. Helping me remember where I left off so that I can dive back in where otherwise I might get stuck overthinking the work I’d already done.

I’m also trying to adapt a daily prompt to keep me accountable to the work. I’ve already created a kind of daily overview prompt and scripts that works towards this goal. The next step is to augment this with a kind of daily standup with an agent.

I can’t imagine ADHD is the only thing that AI can help with, just that this is the most obvious since it effects me. Definitely a wide area and I wonder what others could find useful for their needs.

Wrap Up

AI in general seems to be a spicy umbrella of issues. From mental health to environmental concerns. It’s a wide variety of concerns and unknowns that we seem to be focused on the extremes. A lot of what gets headlines are these wildly polar outcomes, but outside of smaller conversations I have with other devs, no one is talking about how we actually use and are being effective with AI (probably because it’s boring).

There are things I am more excited by with AI than I was a year ago. The big thing is that I think we need to be a bit more practical with it and AI starts to feel more sensible. That’s it.