The biggest problem with LLMs doing code is erosion of expertise. You need to know enough to judge whether an LLM’s code is good, but people who use LLMs for their whole career probably won’t develop expertise, and the people who didn’t learn programming with Claude’s help will retire, probably leaving no one who can tell the quality of the LLM’s output
They’ll never learn that you still have to review the code. The only thing that will absolve you of all work and responsibility is death.
Are you referring to all the people who use AI and still have jobs, or a different group?
Who is “they” in this scenario?
The people who use AI and dont even look what it does. The only explanation for experts using it and this divide growing is they aren’t watching what it does. They’re just letting it do their job for them and not thinking.
So these mythical workers are pushing slop all day every day without reading their output, the entire time AI has existed… and they’re still gainfully employed? And getting promoted?
How is that possible that a worker is producing absolute verifiable garbage, but is still viewed as a valuable member of the team by their peers and supervisors?
Did you read the fucken article? A divide does not mean that they are producing straight garbage, it means that the code quality is lessening, not that it immediately became a mess that won’t even compile. Which in my opinion means nobody gives a fuck about their desk jobs and they still collect a check whether the AI spits out the code or they do.
I’m not talking to an article, I’m talking to you. And you said:
The people who use AI and dont even look what it does.
I disagree with your assertion
Well apparently I am talking to a fucking brickwall
Attacking the person and not the point is probably supporting your argument, somehow
