Ask HN: Are we all walking into a trap? Hello HN, I use AI to code, not all the time but it gets work done. Unless you have been under a rock for the last few years I bet you do to some degree as well. I'm sure we've crossed that line where we won't go back to unaided coding, so.. What are your philosophical thoughts on paying a subscription to get your work done? The new norm? Lets see what local models can do? The richest bird gets the worm? Lets hear, from the humans that are in here please. |
Paying for an IDE, course or fancy tools was paying for complementary additions instead of the core.
Just look at the, often smug, messaging of these companies. Your means of producing economic value are being centralized. The perceived and actual value of software products is being lowered, but some keep bragging about how a bot is doing their job.
Open-source models must win.
That's worrying for any company that depends on these things, and we can already see companies like Amazon and Microsoft freaking out about how much they're spending on AI services for employees to use.
And it's also why I'm hesitant to use them too much. Personally, I refuse to use subscriptions for anything unless absolutely necessary, and I'll always either pick an open source option or a one time purchase if it's available. I think that mentality might become more common in future when these services hike up the price.
Man in the trap wonders if they are walking into a trap.
You've always been doing that. Whether it's called "subscription" or not.
The "you" in this case is yourself, your team or your organization.
Your computer / laptop, the software it runs on, the mouse, keyboard, monitor, desk or whatever else you always had to pay = things you pay to do your work.
Even back in the day you had to pay for pen and paper. Consumables = API credit then?
Point being, what's the difference?
Just like you can code without AI you can also code without a chair. You can hold the laptop and try type.
> The new norm?
No
I think the way to solve this are open weights models, but they are still pretty new and less powerful, from my experience, that SOTA models
I've paid for subscriptions to get work done for years. Educational materials, IDEs, source control tools, debugging tools, etc. I don't think the fact that you're paying for a subscription is the real potential worry here, right? It's more about the effects of using this particular tool.
There have always been costs associated with tools, whether in money or time. A lot of businesses have been paying SaaS subscriptions for decades, and support contracts for longer. I don't understand how AI would be any different.
You are paying to get the work done much quicker. You are saving time, if you don't think your time has more value than the cost of AI agent then you should not pay for these tools.
> The new norm?
As long as companies are ok to pay for these tools and they believe that these do provide value, yes this is going to be the new norm.
> Lets see what local models can do?
My hope is that local models become as good as cloud models but the tricky part is all the AI companies would prefer to have big giant cloud models instead of innovating on optimization so that even small models can do good work.