51–60 of 206 posts
Low-skill work that used to be outsourced will go to cheaper LLMs, unless wages are depressed enough / running costs are high enough to keep using humans as cogs in the machine. This will also consume a ton of small-scale things, like personal-sized automation and small-business customization of better-crafted things (stuff that normally wouldn't be paid for in the first place, or only extremely rarely). Some will ob…
in the olden days (pre-LLMs) we would write high-level code.
the entire layer was high-level code and rarely would we ever need to peak into the assembly:
writing, debugging, architecting, reviewing, testing - all were done in the high-level language layer.
---
welcome to present day:
since we don't write code - we write intents, we also shouldn't review code either - we should review intents.
I don't review my code anymore. I ask the agent to generate markdown docs, graphviz diagrams, changelogs, audit reports, etc. I only review that.
I also ask it to write test and evaluate by whether the tests passed or not. I don't need to peak into the tests code - I can also ask plain english, pseudocode, control flow graph, whatever it is I want.
I can ask it to find errors or missing tests and improve that too!
code is like assembly now.
rare are the cases you would need to peak into that level.
> ask claude to write, and ask claude to explain This works, until it doesn’t. I’m continuously shocked by these stories, where so many people put the future of their job/company in the hands of these agents after only a few months of existing. I still constantly run into bad output from LLMs, from code to basic questions. I don’t understand how anyone can hand things over to something that is laughably wrong on a pr…
There's nothing shocking about this. The vast majority of software/source code is pretty terrible anyways, code that is full of bugs, slow to use, has little to no automated tests and very hard to maintain. To the extent that it gets fixed or works at all, it's not because of competent developers doing rigorous analysis of the software, it's because either someone testing it or using it gets annoyed, reports an issue…
I'm still amazed people don't achieve extremely high test quality, since you get tests "for free" now.
One of the limitations of testing were always that people "design" things so they're hard to test.
And then they argue "This can't be tested", or "Refactoring this for testing is not worth it."
It is now. Yet, I work on codebases with no tests and lots of yolo co-authoring.
I will just say, if you are any good at programming and have experience using agents, you're in the top 0.1% of the world in adoption of a critical new technology. It may seem hopeless as a programmer , but imo you'd be much better off reframing your situation re: the above sentence.
Among peers, I feel like I am top 20%? 30%? maybe, by being a good programmer who is adept at agents. A year ago was the 0.1% point, this stuff is spreading like wildfire. A year from now I think it's going to just be de rigeur that these are our tools now.
Worse, any edge I have from working with this stuff for years is quickly dulled. The tools are evolving fast. My tricks from 3 months ago have been eclipsed.
I'm in a slow-moving, much bigger company. Lot's of talk about "AI" here and we can use copilot if we want to, but there is 0 pressure. I'm in a small team and one colleague uses copilot often. In the beginning there was a minor conflict between him and me, because I found the quality of the LLM code unacceptable and had to ask him to review it more carefully. I think that's settled now, but it makes me sad how a once motivated colleague now seems to try to cheat his way out of work.
I personally find it incredibly boring to write copilot prompts or read its answers full of boiler plate and sycophancy. I don't understand how anyone would want to replace the cognitive work of programming, that I find enjoyable for the most part, with the cognitive work of "talking" to an LLM.
Anyway, I think it will be like this at least for a little while longer: only vibe coding allowed in small companies and less vibe coding the bigger the company is.
But before vibe coding can take over the slow-moving big companies, all the accumulated technical debt will come back to haunt us and vibe-free software will be the new fad. That's what I hope at least.
Earlier quoted context omitted.
There's nothing shocking about this. The vast majority of software/source code is pretty terrible anyways, code that is full of bugs, slow to use, has little to no automated tests and very hard to maintain. To the extent that it gets fixed or works at all, it's not because of competent developers doing rigorous analysis of the software, it's because either someone testing it or using it gets annoyed, reports an issue…
> little to no automated tests I'm still amazed people don't achieve extremely high test quality, since you get tests "for free" now. One of the limitations of testing were always that people "design" things so they're hard to test. And then they argue "This can't be tested", or "Refactoring this for testing is not worth it." It is now. Yet, I work on codebases with no tests and lots of yolo co-authoring.
> ask claude to write, and ask claude to explain This works, until it doesn’t. I’m continuously shocked by these stories, where so many people put the future of their job/company in the hands of these agents after only a few months of existing. I still constantly run into bad output from LLMs, from code to basic questions. I don’t understand how anyone can hand things over to something that is laughably wrong on a pr…
One possibility is that software starts to look more like traditional manufacturing.
The machine is the company’s core asset. The engineer only needs to know how to operate the machine well. Once that happens, the barrier gets much lower, need much less people, and the job naturally become much less valuable. Some parts will still need to be done by hand, of course. But only a very small part. It is like old factories. They used to need lots of fitters, at all levels of skill. Now you only need a few of the elite ones.
AI is the CNC machine of the software industry.
The more pessimistic future is that, maybe five years from now, the best programmers will look at AI the same way the best Go or chess players look at AI today: Like KeJie said, "I don't even know what I am trying so hard against." We now have a new SOTA every two months. It just took 18 months for LLMs from reasoning models to disproving the unit distance conjecture. ChatGPT itself has not even existed for as long as a college student spends in university.
In any case, we have already passed the point where this can be rolled back.
Maybe ten years from now I will be leaving a comment saying that "programmer" used to be a job too :-/
Programming is the low-hanging fruit for AI. Open source and knowledge sharing have given it huge amount of public, high-quality training data at a level other industries can hardly imagine. And almost everything in programming can be tested and verified inside the computer quickly in a closed loop. No robot arm is needed.
The main weakness of current LLMs is still that they are static: They do not really change themselves through use. Harness tools are just elaborate ornamentation on top of prompts. LLMs are frozen at the moment training stops. Once we get models that can change their own weights through self-feedback, then maybe AGI really is on the horizon.
Thinking optimistically: I may be lucky enough to see it in my lifetime. Maybe by then, people will be able to live more like human beings, instead of organizing their whole lives around work :-)