It’s copium to think that LLM code is more buggy than your median to slightly above median developer or that’s all that most companies need - median developers.
And debugging code is also easier with AI. Just today I had to revisit code that I personally wrote from the design, the implementation the refactoring, etc from the first git init and I couldn’t remember half the decisions I made. I launched Codex and started asking it questions about the code.
Where is the productivity gain? How many junior developers and mid level ticket takers are struggling to find a job now because the market is saturated and those true seniors who can operate at a larger scope and impact can do the work themselves without having to delegate
My personal anecdote is that I had four offers within 3 weeks after being Amazoned in late 2023. One was from the company that acquired the startup I left in 2020 where I would have been responsible for leading the integration between all of the companies they acquired [1] and the other was a former coworker who was now a director st a well known non tech F500 company. He wanted me to lead the migration and “modernization “ efforts. I decided to stick with consulting.
Those offers didn’t come because of my coding abilities. That’s a commodity.
I was looking again in 2024. It took one outreach and talking to the right people. Absolutely no one asked me the first thing about coding even though I do it maybe 60%-70% of the time.
Going way back to 2016, I had two offers - one interview was me doing a merge sort on the whiteboard the other interview was me talking about strategy with the then new director who needed to build up a software development team. He asked me about my experience. He didn’t mske me stand up and do some algorithmic test on the whiteboard. He treated me like an industry professional
[1] I did the whole “lead integration efforts by a company owned by private equity acquiring other companies” thing before I joined the startup - never again.
> It’s copium to think that LLM code is more buggy than your median to slightly above median developer or that’s all that most companies need - median developers.
You are not understanding the point. AI has to be properly supervised because it makes mistakes. Now if you are making more or as many mistakes as the AI, then you should look for a new career. You should understand the code better than the AI, because the AI has a limited context window, and for a large codebase, you should know that codebase better than the AI.
Now, you can use AI to help you understand code that someone else wrote. You can use AI to check your code. You can use AI to write unit tests. You can use AI to debug. You can use AI to summarize code. There are so many uses of AI.
But you -- you as a developer -- need to understand your codebase. If you do not understand the codebase, you can't properly supervise the AI. And there is one efficient way for you to understand a big complicated codebase. The most efficient way possible for you to learn it. That is by you writing code in that codebase, and debugging that code, and learning how to code in that codebase.
If you don't do that, then you are not qualified to supervise the AI. Now you are letting the AI loose on a codebase much larger than its context window, and you will fill the codebase with bugs.
It's like a student. Sure you can use AI to help you study, to explain things to you, but the moment you let the AI do your homework, then you are no longer learning. The homework is the practice of solving problems in that area, and as a developer, you need to write code in a codebase otherwise you have no value and the AI has no value.
> You are not understanding the point. AI has to be properly supervised because it makes mistakes.
It been well over a decade (ironically until AI) that I have been responsible for an implementation small enough that I could do with my own two hands within the allotted time by myself. People also require supervision.
Now if you are making more or as many mistakes as the AI, then you should look for a new career.
Right now in 2026, the code I write is dependent on the AWS SDK.
It’s the same surface area for every supported language - as new features come out AWS uses code gen tools to update the SDK and CLI based on a common set of specifications.
Consider it’s the same surface area as when I have to write IAC. Are you saying it should be faster for me to know that API and write it by hand than an LLM? Especially now that AWS has an MCP server that AI can use to know the latest documentation. The case was the same in 2000 when I had to write C and C++ against the Win32 APIs, DCOM etc or when I didn’t know the entire C++ STL and had to reference the “C++ programming language ”.
> You should understand the code better than the AI, because the AI has a limited context window, and for a large codebase, you should know that codebase better than the AI.
Every developer should know the entire codebase even when they first join a company? Even when I was working at a 70 person startup, that had three teams not one person knew the entire codebase.
> But you -- you as a developer -- need to understand your codebase. If you do not understand the codebase, you can't properly supervise the AI.
Again how large of a codebase is a developer suppose to know? S3 for instance is made up of 300 separate microservices. Is a single developer suppose to know how it all works?
Besides that’s why we have unit tests and should be writing modular code so you don’t have to worry about spooky action at a distance.
And people are so focused on “the code”. I have designed systems and architectures that include code and multiple teams and consulting companies. Am I suppose to know how all of the code works?
> And there is one efficient way for you to understand a big complicated codebase. The most efficient way possible for you to learn it. That is by you writing code in that codebase, and debugging that code, and learning how to code in that codebase.
Before AI, when coming into a team, I didn’t just start “coding” I would ask other people. Now I ask AI.
Are you really saying that you can keep more of the code in your head than AI with a one million token context window or that you can read and understand various markdown files in a repo faster?
> Are you really saying that you can keep more of the code in your head than AI with a one million token context window or that you can read and understand various markdown files in a repo faster?
Of course. A million tokens is 50,000 lines of code. That's nothing.
It's not that you memorize a million lines of code. It's that you develop a solid mental model of how those lines of code are organized, how they work together, what the important gotchas are, what hurt performance in the past. You know the system well. Then when you are interacting with something, you can generally make predictions about the best way to make a change to that system, and you know in which modules to look when diagnosing a problem, so you can work your way around the codebase efficiently.
Most large mature projects are millions of lines of code. Chrome, Photoshop, databases like oracle or postgres.
But the LLM with that million token context window has that 50,000 line memory and then its general training data. That's it.
So your "value" as a developer who is using AI is to bring that mental model with you so you can supervise the AI and guide it to do surgery on that million line codebase with it's puny 50,000 line knowledge of your codebase.
But when you stop coding, you begin to lose your mental model. So, yes, there is a short term burst of productivity as your mental model fades but you are still good enough to guide the LLM. But after a while -- and depending on how deeply that code has been burned into your mind, it could be 6 months or a couple of years -- you will no longer be able to effectively supervise the LLM. You'll be like that noob that wanders into a big codebase they don't understand anymore. Then you wont be able to supervise the LLM.
You have to keep coding in a codebase to maintain your understanding of it. It really is just like having the LLM do your homework for you. Use the LLM to help you learn, to help you do your work, but keep doing the exercises yourself, so that your mental model remains fresh.
> It's not that you memorize a million lines of code. It's that you develop a solid mental model of how those lines of code are organized
As you do when you are leading a team or are more concerned with the overall system - people, business processes, architecture, etc.
I don’t need to write every single if statement to do that
> But the LLM with that million token context window has that 50,000 line memory and then its general training data. That's it.
Absolutely no person us keeping the entire 50K lines of code in memory at the same time - they have a mental model of how the components interact. You don’t lose that mental model when you use LLMs.
> So your "value" as a developer who is using AI is to bring that mental model with you so you can supervise the AI and guide it to do surgery on that million line codebase with it's puny 50,000 line knowledge of your codebase.
Isn’t that what I’ve been saying the entire time?
> But when you stop coding, you begin to lose your mental model. So, yes, there is a short term burst of productivity as your mental model fades
Is you scope of responsibility only what you code yourself? Before LLMs and before I got into consulting, I was responsible for the architecture of a startup with three teams and I did my own MVPs to prove out concepts so I wouldn’t be an “architect astronaught”. I went months with out a line of code. But I did understand the system.
Absolutely no one is paying me to write for loops or understand every line of code. I get paid …decently…for understanding systems. The business, the personalities, conflicting priorities, teasing out the level of technical maturity of the customer, designing the architecture, addressing cross cutting concerns, the modularity of the code etc. the code is no more important that I do it by hand today than it is for me to be writing assembly (or even c) instead of using a compiler.
It’s also not important I know how AWS procures hardware when I submit a YAML file like it was when I was partially responsible for an on prem server room with a whopping 3TB SAN in 2004.
My “mental model” is based on again the processes I’ve perfected to go from requirements of a large implementation -> happy paying customer from an empty git repo and an empty AWS.
The “code” is a minuscule part of any large project. In most enterprise companies that can be outsourced to lower paid people in non US countries. There is no moat around “I know how the for loops work”. For now there is a moat to being able to fly out to a customer site and work with “the business” or on a Zoom call and being able to understand systems. The people who pay you to help you make money or save them money don’t care
But I better be able to explain to the PMO where the project stands, to legal whether we are meeting our contractual obligations, to sales/account management whether the customer is happy and will pay us.
On the customer side, I need to be able to intelligently answer questions that come my way from the security folks, the DevOps department, the bean counters and the development department who have to maintain the project.
Do I need to understand the why and how the system interacts and the behavior of the system? Of course, do I need to know how every for loop was written instead of a while loop? No, I don’t know that if I have a team under me or if I’m responsible for validating the Salesforce integration we outsourced (when I was at a startup).
Coding - in enterprise dev - has been an undifferentiated commodity where the comp hasn’t kept up with inflation in a decade. I saw this happening and was the reason I aggressively moved up the stack as far as scope and responsibility. With LLMs I can take on much larger scoped projects by myself and get them done in 40 hours
> You have to keep coding in a codebase to maintain your understanding of it. It really is just like having the LLM do your homework for you. Use the LLM to help you learn, to help you do your work, but keep doing the exercises yourself, so that your mental model remains fresh.
I graduated from college in 1996. I’ve been assigning other people to do a large part of my “homework” for over a decade. That could be other in house developers or another company to do things like integrations with Salesforce, Workday etc. now it’s Claude and Codex.
And debugging code is also easier with AI. Just today I had to revisit code that I personally wrote from the design, the implementation the refactoring, etc from the first git init and I couldn’t remember half the decisions I made. I launched Codex and started asking it questions about the code.
Where is the productivity gain? How many junior developers and mid level ticket takers are struggling to find a job now because the market is saturated and those true seniors who can operate at a larger scope and impact can do the work themselves without having to delegate
My personal anecdote is that I had four offers within 3 weeks after being Amazoned in late 2023. One was from the company that acquired the startup I left in 2020 where I would have been responsible for leading the integration between all of the companies they acquired [1] and the other was a former coworker who was now a director st a well known non tech F500 company. He wanted me to lead the migration and “modernization “ efforts. I decided to stick with consulting.
Those offers didn’t come because of my coding abilities. That’s a commodity.
I was looking again in 2024. It took one outreach and talking to the right people. Absolutely no one asked me the first thing about coding even though I do it maybe 60%-70% of the time.
Going way back to 2016, I had two offers - one interview was me doing a merge sort on the whiteboard the other interview was me talking about strategy with the then new director who needed to build up a software development team. He asked me about my experience. He didn’t mske me stand up and do some algorithmic test on the whiteboard. He treated me like an industry professional
[1] I did the whole “lead integration efforts by a company owned by private equity acquiring other companies” thing before I joined the startup - never again.