hperrin comments on Alternative to Google Tasks, Calendar, Keep, and some more

[–] hperrin@lemmy.ca 7 points 3 weeks ago* (last edited 3 weeks ago) (4 children)

I feel like you don’t understand why this is important.

You generated the app using subsidized AI. You probably don’t have the hardware to run frontier models on your own, so you’re probably using Claude, which is why I say it’s subsidized. Anthropic doesn’t charge you the real cost of running the AI, let alone enough to actually make a profit off your use of it. That’s fine for now, because they’re swimming in VC money. They can afford to get you hooked by subsidizing your use.

What happens when the VC money dries up? Well, Anthropic is going to have to charge you enough to make a profit. That’s probably going to mean that what costs you $20 today is going to cost you at least $200 then. Are you willing to pay ten times what you’re paying now to have Claude work on this app for you? If you are, then no problem. But if you aren’t, that means you’re probably going to abandon this project. I doubt you want to clean up this code base and work on it without the AI.

If you tell me that this is AI generated, I’m going to know that the risk that you’ll abandon it is much higher than if you are working on it yourself. So, I’m going to be able to make a much better risk assessment about pouring my time, resources, and data into using it. That’s why it’s important. Yeah, it sucks that you’ll get way less adoption telling people the truth, but if growing your users is more important to you than being honest, I wouldn’t want to use your product even if you did write it yourself.

permalink

fedilink

source

parent

hideshow 4 child comments

[–] chedieck@lemmy.ml [S] 2 points 3 weeks ago (3 children)

You are correct about almost everything, except that I'd abandon the project. I had been already a software engineer for four years, before AI came along, and I really enjoyed. I agree with you that this party is not gonna last forever, but I still think makes total sense to take advantage of it while it lasts. I don't ask Claude to keep adding features like crazy, I ask it to create specific tests, I ask him to recreate some parts of the codebase using shared components, I ask him to split services from API routes, etc. I try to keep it organized — sure, I am not auditing every of it, because I simply can't, but this is not messy slop. How do I know? Because I've been working HARD for it NOT to be slop, for more than six months.

If you tell me that this is AI generated, I’m going to know that the risk that you’ll abandon it is much higher than if you are working on it yourself

It's not AI generated. From the downvotes I got I see that people mostly don't understand this, but there is huge difference between prompting "Make me a webapp that has tasks, a calendar, todo lists, goals and everything else that's important and make no mistakes so it works well" and from, for the span of six months, carefully thinking about which stack to use, how to structure data, how to present it in forms and readable tables so the UX is not shitty, etc.

I think the FOSS community right now is at some kind of boy-cries-wolf situation where the spam of AI slop is so great that many people are willing to throw the baby out together with the bath water when they see a project that is not slop but that was created with the help of AI:

"Oh look, OP used AI in his project instead of writing code like a good Neanderthal, let's shit on his head"

The truth is we'll all have to adapt. If I, completely by myself, managed to do what I think that is the best life management app out there (yes I am biased) — what could people do together? Maybe the time has come where every proprietary solution has a good, and I mean GOOD, alternative. We can do that, using AI, but surely not if we treat it with all this hostility because of all the evil associated with it — I said I was a software engineer for 4 years before AI came along, you guys think I don't miss it? When I easily felt valued, because I could use my brain to solve REAL problems in the REAL WORLD?

Well, that time is over. Cry now, cry later, or don't cry at all and try to do the best with the shit timeline we're stuck in.

permalink

fedilink

source

parent

hideshow 3 child comments

[–] hperrin@lemmy.ca 3 points 3 weeks ago* (last edited 3 weeks ago) (1 child)

I see that you added a note at the top about using LLMs. Thank you. :) I feel much better about this project now, and if you post it again, I hope you do get a much better response.

permalink

fedilink

source

parent

hideshow 1 child comment

[–] chedieck@lemmy.ml [S] 1 point 3 weeks ago

I got tired of all the back and forth, but you guys were definitely right about the transparency of having a note explicitly saying what I am doing making things generally better

permalink

fedilink

source

parent

[–] hperrin@lemmy.ca 2 points 3 weeks ago* (last edited 3 weeks ago)

You’re getting downvoted because you didn’t disclose that it was built with AI, then you got defensive when you got called out on it. You can think we all hate AI all you want, but what we really hate is dishonesty.

Your commits don’t list Claude as an author, they list you. But you didn’t actually write those commits. You claim copyright ownership over this codebase, too. Maybe you’re unaware, but you can’t actually copyright something that was AI generated. At best, you own copyrights to some of the code.

Congratulations on being a dev for four years before AI. Are you really willing to ruin your reputation now by not being truthful about where the code in your codebase came from?

Also, you couldn’t even delete the files that Claude left in there yourself. What possible indication do I have that you would work on this project without Claude there to do everything for you?

permalink

fedilink

source

parent