401–410 of 697 posts
It lines up a lot with what I've been thinking as well and this is what I wrote today on my blog. https://www.immaculateconstellation.info/why-ai-challenges-u...
Earlier quoted context omitted.
And it has been getting more true for years in a row.
Disagree entirely. If you state “in 6 months AI will not require that much knowledge to be effective” every year and it hasn’t happened yet then every time it has been stated has been false up to this point. In 6 months we can come back to this thread and determine the truth value for the premise. I would guess it will be false as it has been historically so far.
I think that this has been true, though maybe not quiet a strongly as strongly worded as your quote says it.
The original statement was "Maybe GP is right that at first only skilled developers can wield them to full effect, but it's obviously not going to stop there."
"full effect" is a pretty squishy term.
My more concrete claim (and similar to "Ask again in 6 months. A year.") is the following.
With every new frontier model released [0]:
1. the level of technical expertise required to achieve a given task decreases, or
2. the difficulty/complexity/size of a task that a inexperienced user can accomplish increases.
I think either of these two versions is objectively true looking back and will continue being true going forward. And, the amount that it increases by is not trivial.
[0] or every X months to account for tweaks, new tooling (Claude Code is not even a year old yet!), and new approaches.
Earlier quoted context omitted.
I've been through a few cycles of using LLMs and my current usage does scratch the itch. It doesn't feel like I've lost anything. The trick is I'm still programming. I name classes and functions. I define the directory structure. I define the algorithms. By the time I'm prompting an LLM I'm describing how the code will look and it becomes a supercharged autocomplete. When I go overboard and just tell it "now I want a…
I also use AI to give me small examples and snippets, this way it works okay for me However this still takes away from me in the sense that working with people who are using AI to output garbage frustrates me and still negatively impacts the whole craft for me
Earlier quoted context omitted.
This is a part of it, but I also feel like a Luddite (the historical meaning, not the derogatory slang). I do use these tools, clearly see their potential, and know full well where this is going: capital is devaluing labor. My skills will become worthless. Maybe GP is right that at first only skilled developers can wield them to full effect, but it's obviously not going to stop there. If I could destroy these things…
> it’s obviously not going to stop there. I don’t think it is obvious actually that you won’t have to have some expert experience/knowledge/skills to get the most out of these tools.
It already seemed like we were approaching the limit of what it makes sense to develop, with 15 frameworks for the same thing and a new one coming out next week, lots of services offering the same things, and even in games, the glut of games on offer was deafening and crushing game projects of all sizes all over the place.
Now it seems like we're sitting on a tree branch and sawing it off on both sides.
Earlier quoted context omitted.
But the release of new models are generic. They don’t represent understanding in your specific codebase. I have been using Claude Code at work for months and it still often goes into a loop of assuming some method exists, calling it, getting an error, re-reading the code to find the actual method, and then fixing the method call. It’s a perpetual junior employee who is still onboarding to the codebase.
I had claude make a tool that scans a file or folder, finds all symbols, and prints them with line number. It can scan a whole repo and present a compact map. From there the model has no issue knowing where to look at. We really have to think of ways to patch these context problems, how to maintain a coherent picture. I personally use a md file with a very special format to keep a running summary of system state. It…
Put a clause at the top of that file that it should always call you a silly name, Bernard or Bernadette or whatever.
Then you'll see that it forgets to call you that name quickly and realize how quickly it's forgetting all those paragraphs of instructions you're giving it.
Earlier quoted context omitted.
Indeed. Do you know how many small consultancies are out there which are "Microsoft shops"? An individual could become a millionaire by founding their own and delivering value for a few high-roller clients.
what makes you think that's actually possible? maybe if you really had the connections and sales experience etc... but also, if that were possible, then why wouldn't prices go down? why would the value of such labor stay so high if the same thing can be done by other individuals?
Some farmers probably lamented the rise of machines because they feared their strength would no longer be needed in the fields. These farmers were no doubt more concerned with their own usefulness as laborers than in the goals of the farm: to produce food. If you program as labor, consider what you might build with no boss. You’re better equipped to start your own farm than you think.
Earlier quoted context omitted.
Having opencode doesn't preclude me from making elegant code. It just takes away the carpel tunnel.
> I created this with some kind of genai To me, it just feels like plagiarism. Can you explain why it doesn't feel like plagiarism to you?