Joel, It's a damn good question, as I am reading this in between copy and pasting prompts from Claude into Lovable and then results from lovable back into Claude as the tools implement a strategy I worked with Claude to develop for my new EchoBBS.com hobby platform to bring back the nostalgic BBSs to the web. I know what I want and they are implementing my shared strategy I honed against Claude to perfect, but I don't really understand how it is all being built.
The truth is, as a product manager, I did the same thing. Define what I wanted and let the engineers build it. I could always ask questions and diagnose solutions, but never really knew how anything was really built and working on the back end. AI just makes the work faster and allows me to turn ideas into reality that I would have never been able to do by myself in the past.
I think it's different. AI allows me to do things I couldn't do otherwise. That's not a loss of capability. It's a net gain. But if I source all the things I already am competent in, that's a net loss.
100% - I could see how easily you could get drawn in to letting AI do everything. It's getting smarter and smarter every day. Knowing what you are great at and using AI to fill in the gaps is the key.
I get your point, Adam, but there is also a big difference between a hobby project and a commercial product people paid for and rely on. I did a hard no on a guy who was telling his marketing mentorees to vibe code products to sell. He spent zero time on testing--didn't even know if anything really worked--and didn't teach how to support or fix the very things he was proposing people sell.
And don't forget even when you were a product manager, the engineers knew how things were built. The innards were not your responsibility, but someone was still on the hook.
And yes, A.I. coding tools can troubleshoot and fix bugs... sometimes. And when they do they are not afraid to gut the project; unwanted side effects abound. I have also got into a loop with Claude where his fix broke many things, and then we went on and on with fix after fix, only to reintroduce the original bug on what he was sure was the last iteration. That was fun.
But you are correct, A.I. tools do allow us to do things we were not able to do in the past. No question about that. I love them, too. As the sergeant used to say on Hill Street Blues (yep, I'm old), "Be careful out there."
LOL, I have to admit, I have, too. But... if your AI tool gets stuck and you're not a coder, what do you do?
I just think there's a lot more to creating a software product--and the company that supports it--than vibe coding a bunch of features in an afternoon.
There are several books I wish all of humanity was required to read (and yes, the Bible is #1). But as time goes on, this little short story is rapidly climbing towards the #2 position:
Joel, It's a damn good question, as I am reading this in between copy and pasting prompts from Claude into Lovable and then results from lovable back into Claude as the tools implement a strategy I worked with Claude to develop for my new EchoBBS.com hobby platform to bring back the nostalgic BBSs to the web. I know what I want and they are implementing my shared strategy I honed against Claude to perfect, but I don't really understand how it is all being built.
The truth is, as a product manager, I did the same thing. Define what I wanted and let the engineers build it. I could always ask questions and diagnose solutions, but never really knew how anything was really built and working on the back end. AI just makes the work faster and allows me to turn ideas into reality that I would have never been able to do by myself in the past.
I'm a big fan.
Adam
I think it's different. AI allows me to do things I couldn't do otherwise. That's not a loss of capability. It's a net gain. But if I source all the things I already am competent in, that's a net loss.
100% - I could see how easily you could get drawn in to letting AI do everything. It's getting smarter and smarter every day. Knowing what you are great at and using AI to fill in the gaps is the key.
I get your point, Adam, but there is also a big difference between a hobby project and a commercial product people paid for and rely on. I did a hard no on a guy who was telling his marketing mentorees to vibe code products to sell. He spent zero time on testing--didn't even know if anything really worked--and didn't teach how to support or fix the very things he was proposing people sell.
And don't forget even when you were a product manager, the engineers knew how things were built. The innards were not your responsibility, but someone was still on the hook.
And yes, A.I. coding tools can troubleshoot and fix bugs... sometimes. And when they do they are not afraid to gut the project; unwanted side effects abound. I have also got into a loop with Claude where his fix broke many things, and then we went on and on with fix after fix, only to reintroduce the original bug on what he was sure was the last iteration. That was fun.
But you are correct, A.I. tools do allow us to do things we were not able to do in the past. No question about that. I love them, too. As the sergeant used to say on Hill Street Blues (yep, I'm old), "Be careful out there."
Ross, I have had that same experience with human engineering teams. :-)
LOL, I have to admit, I have, too. But... if your AI tool gets stuck and you're not a coder, what do you do?
I just think there's a lot more to creating a software product--and the company that supports it--than vibe coding a bunch of features in an afternoon.
There are several books I wish all of humanity was required to read (and yes, the Bible is #1). But as time goes on, this little short story is rapidly climbing towards the #2 position:
https://www.cs.ucdavis.edu/~koehl/Teaching/ECS188/PDF_files/Machine_stops.pdf
It is "The Machine Stops", a short story by E.M. Forster written around 1909. The dude was a prophet. It speaks directly to your musings, Joel.
P.S. I now confess the truth: I have a degree in English with a focus on literature. I came to technology somewhat by accident.
Read that short story people! Forster was uncannily prescient and 100 years ahead of his time.