We Are Not Going to Agree on AI

On one hand, I know a developer producing 30,000 lines of code a month. On the other, I know a developer who says AI is stupid. Each swears by their stance and has evidence to back it up. One has a working product and the other has a broken one.

The New York Times profiled Medvi and reported they're on track to make $1.8 billion this year. Clearly AI worked for them... if you ignore the alleged fraud for just a second. And while Microsoft now claims that at least 30% of their code is AI-generated, GitHub logged 89 incidents in 90 days (as of this writing). That doesn't exactly paint a bright picture of a technology firing on all cylinders.

If you're sitting on the sidelines trying to decide whether AI works or not, you're not going to get a clean answer. But it's still the right question to be asking.

I don't think we'll ever reach a consensus, because after all the hype, what we're left with is a capable tool. And apparently, that's not enough. For AI companies, it's supposed to be the alpha and the omega. Something that will both kill us and save us, take all our jobs and liberate us at the same time. For the rest of us, if we're not afraid, it's proof we don't understand it well enough, and we'll be left behind.

I think of AI as a capable tool. That's it.

I went to Home Depot once to buy what I needed to mount a TV. I asked an employee for a stud finder, and instead of just pointing me to the aisle, he walked me there himself. I hate when they do that, it usually means they're about to try to sell you something.

Sure enough, in Aisle 17, his coworker was manning a caged shelf stocked with expensive-looking tools. When the cage opened, he didn't reach for the simple one I wanted. He grabbed the model loaded with 13 additional sensors. I asked if the basic one could do the job. The first employee took my side and made the case for simplicity. The second shot back: "Sure, if you don't mind drilling through a live wire!"

I stood there watching these two argue, trading field stories like they were one-upping each other at a bar. One of them claimed he'd worked with a guy who didn't even need a stud finder, he'd just knock on the wall three times and know exactly where to drill. I put both stud finders in my basket, thanked them both, and walked away.

I circled the store a few times to make sure neither of them could see me before I quietly dropped the expensive one into an empty basket near the checkout. To this day, I'm a little afraid to go back to that Home Depot.

The only thing that matters to me is what I can do with a tool, not what the tool can theoretically do. When a tooltip pops up on my screen, I dismiss it before I've even finished reading it. I don't care about Jira's latest feature update on the sidebar. I don't care that AI can rewrite my already written ticket. I just want the stud finder to help me hang the TV. Everything else it can do holds no interest for me, especially when I'll use it for ten minutes and not touch it again for three years.

When I have a goal, I reach for whatever helps me get there. If you're waiting for a tool to do the work for you, you're going to be disappointed. A tool's job is to make your work easier, and sometimes it does, sometimes it doesn't. Figuring out when to reach for it is on you.

---