Hacker News Reader

Comment by chii | Hacker News Reader

> A much larger percentage will only be able to work on things like CRUD apps and grunt work.

which is lower valued, and thus it is economically "correct" to have them be replaced when an appropriate automation method is found.

> When these jobs are eliminated, many developers will be out of work.

like in the past, those developers will need to either move up the value chain, or move out into a different career. This has happened before, and will continue to happen until the day humanity reaches some sort of singularity or post-scarcity.

View on HN

> This has happened before

When do you think this is most comparable to?

Were there this many software developers around the peak of the dot com era? Im 35 so im old enfough to remember the excess of the time and all the weird products, but nothing about how the sausage was being made.

> When do you think this is most comparable to?

I don't think it has happened to the software industry. We've been growing as computers have become more and more capable since the 1960s.

One analogy is farm employment cratering in the first half of the 20th century.

Went from 11.77M in 1910 to 5.88M in 1950 in the USA[0], even as the population went from 92M to 151M[1]. That's going from 12% of the population to 4%.

Another is travel agents, which went from 100k workers in the USA in 2001 to 30k in 2020[2] (though there has been a rebound since).

0: https://ourworldindata.org/employment-in-agriculture#all-cha...

1: https://www.census.gov/data/tables/time-series/dec/popchange...

2: https://fred.stlouisfed.org/series/LEU0254497900A

IMO it happened a bit more than ten years ago when people who were making websites had to specialize between frontend development and UI/UX going from a larger web designer/developer role.

Here's an article talking about it https://css-tricks.com/the-great-divide/

> which is lower valued, and thus it is economically "correct" to have them be replaced when an appropriate automation method is found.

Textbook example of why this “economic” form of analysis is naive, stupid, and short-sighted, (as is almost always the case).

AI models will never obtain the ability to completely replace “low value work” (as that is not perfectly definable or able to be defined in advance for all cases), so in a scenario where all engineers devoted to these tasks are let go, what you would end up with is a engineers higher up the value chain being tasked with resolving the problems that result from when the AI fails, underperforms, or the assessment of a task’s value was incorrect. The cumulative effect of this would be a massive drain on the effectiveness of said engineers, as they’re now tasked with context switching from creative, high-value work to troubleshooting opaque, AI code slop.

> AI models will never obtain the ability to completely replace “low value work”

if this were truly the case, then companies that _didn't_ replace the "low value work" by ai and continued to use people will outperform and outcompete. My prediction is entirely predicated on the ability for the LLM to do the replacement.

A second alternative would be that the cost of the "sloppy" ai code is externalized, which is not ideal but the past history has any bearing, externalization of costs is rampant in corporate profit struggles.

> AI models will never obtain the ability to completely replace “low value work"

Maybe, but this is not the meaning of replacement in this context and it need not hold for the "economic" reasoning to work.

All that matters is that AI makes developers more productive, as measured by number of (CRUD or whatever) apps per developer per unit of time. If this is true, then the current supply of apps can be provided by fewer developers, meaning that some of the current developers aren't needed anymore to sustain the current production level. In this scenario lower level engineers still exist, they are just able to do more in the same time by using AI.

I’m working on a system now where the hard part is the integration, user experience, business requirements, solving XYProblems, etc.

Honestly this is true for most problems and has been forever for most developers.

But between all of the different Lambdas (yes we had to use Lambda to do business logic it’s Amazon Connect), there is probably around 2000 lines of relatively straightforward code and around 1000 lines of infrastructure as code.

I didn’t write a single line of code, I started by giving ChatGPT the diagram and very much did “vibe coding” between the code, the database design and the IAC.

I would have had to have at least one junior dev do the grunt work for me. I don’t think I wrote a single line of code.

Before the gate keeping starts, I started programming in assembly in 1986 and had an official title of “software engineer” or something similar until 2020.

Reply on news.ycombinator.com