The hype has kinda died down for me. Like you said it's a different beast. Vibe coding is fun for prototyping and MVPs, but it ultimately kinda slows me down in the long run. It really reminds m of circa-2012 when I would build clickable prototypes in InVision and think that the future was incredible. It became a good part of most flows but it didn't replace coders or anything.
I'm in the hype cycle for sure. I can also see how it would slow down in the long run, but the thing that gets me is that it slows down for the same reason human programmers slow down. The accumulation of code, of implementation patterns and implicit requirements that were hacked in, discovered along the way, etc.
e.g. it's not slowed down because of something inherent in AI.
As someone with considerable experience with software development, you can effectively communicate your requirements to the AI, spot its mistakes, debug the communication gap, and try again.
Folks just entering the field lack that experience. AI replaces many of the tasks required to gain that experience. Writing a bunch of low-quality code and having a senior engineer critique it is a rite of passage that leads to improvement.
Do you see a danger in companies neglecting to invest in new talent since it may be more effective and certainly much cheaper to have a few seniors with AI superpowers than a few seniors with a team of juniors?
Will tech giants be the only ones with the resources to train new talent?
How do you see all this playing out over the next few years?
I don't disagree with your premise. Not one bit. It's scary.
I don't know how it plays out. There's an incentive problem.
Have you read "Slack" by Tom DeMarco? One of the core ideas in that book is that software developers aren't fungible, and treating them like they are drives them away and is a one-way train to shit product quality town. Same with most other knowledge work. And one consequence of this is that companies need to figure out how to leverage the superpower of the team members they do have. And to seek out the best.
But let's face it - the global marketplace is competitive, and most companies just cannot, and will not, get the best talent. For the best companies, maybe some of them will continue to invest in the most promising young people.
For the rest of them?
I also think it's worth considering the profit margins of various business models as well. For venture scale saas - basically giant margins. But for non-saas companies, or more traditional companies that have brought development in-house....if their margins are tight, guess which category of headcount represents the highest cost? Just imagine the conversation with the CFO that happens behind closed doors at those companies.
I think a major consequence of all this will be that the best engineers will be fine, especially if they're able to leverage the new paradigm. Everyone actually gets a superpower. But how will the less-than-best engineers fare?
Last thought. Whenever there's some new industrial automation, what often happens is that everyone expects the workers to get totally fucked. But instead, the automation drives the cost down to such a degree that demand for talent actually increases instead. There's a word for this, I just can't remember what it's called. That could definitely happen here too.
But yes, if I am someone brand new to software development, candidly I would have a great amount of trepidation unless I was already on some fast-track to a great career, like an internship at an established SV tech co or a co-op at a great school that those companies recruit from, like Waterloo.
The hype has kinda died down for me. Like you said it's a different beast. Vibe coding is fun for prototyping and MVPs, but it ultimately kinda slows me down in the long run. It really reminds m of circa-2012 when I would build clickable prototypes in InVision and think that the future was incredible. It became a good part of most flows but it didn't replace coders or anything.
I'm in the hype cycle for sure. I can also see how it would slow down in the long run, but the thing that gets me is that it slows down for the same reason human programmers slow down. The accumulation of code, of implementation patterns and implicit requirements that were hacked in, discovered along the way, etc.
e.g. it's not slowed down because of something inherent in AI.
As someone with considerable experience with software development, you can effectively communicate your requirements to the AI, spot its mistakes, debug the communication gap, and try again.
Folks just entering the field lack that experience. AI replaces many of the tasks required to gain that experience. Writing a bunch of low-quality code and having a senior engineer critique it is a rite of passage that leads to improvement.
Do you see a danger in companies neglecting to invest in new talent since it may be more effective and certainly much cheaper to have a few seniors with AI superpowers than a few seniors with a team of juniors?
Will tech giants be the only ones with the resources to train new talent?
How do you see all this playing out over the next few years?
It's Jevon's Paradox: https://en.wikipedia.org/wiki/Jevons_paradox
I don't disagree with your premise. Not one bit. It's scary.
I don't know how it plays out. There's an incentive problem.
Have you read "Slack" by Tom DeMarco? One of the core ideas in that book is that software developers aren't fungible, and treating them like they are drives them away and is a one-way train to shit product quality town. Same with most other knowledge work. And one consequence of this is that companies need to figure out how to leverage the superpower of the team members they do have. And to seek out the best.
But let's face it - the global marketplace is competitive, and most companies just cannot, and will not, get the best talent. For the best companies, maybe some of them will continue to invest in the most promising young people.
For the rest of them?
I also think it's worth considering the profit margins of various business models as well. For venture scale saas - basically giant margins. But for non-saas companies, or more traditional companies that have brought development in-house....if their margins are tight, guess which category of headcount represents the highest cost? Just imagine the conversation with the CFO that happens behind closed doors at those companies.
I think a major consequence of all this will be that the best engineers will be fine, especially if they're able to leverage the new paradigm. Everyone actually gets a superpower. But how will the less-than-best engineers fare?
Last thought. Whenever there's some new industrial automation, what often happens is that everyone expects the workers to get totally fucked. But instead, the automation drives the cost down to such a degree that demand for talent actually increases instead. There's a word for this, I just can't remember what it's called. That could definitely happen here too.
But yes, if I am someone brand new to software development, candidly I would have a great amount of trepidation unless I was already on some fast-track to a great career, like an internship at an established SV tech co or a co-op at a great school that those companies recruit from, like Waterloo.
What do you think?