There's a divide between Engineering a system, and implementing it in most other fields. We're now facing the situation where LLMs have artificially accellerated the trend we were already on, towards a professionalization of the art of Software Engineering. The title inflation of most of the programing jobs being incorrectly labeled as "software engineer" has obscured this trend.
The demand for actual Engineering skills won't go away anytime soon. The thing about Engineers is accountability and responsibility. If you're responsible for outcomes, you must exercise authority over the process. This gives you bargaining power for wages, which FAANG and others don't want you to have.
The delineation between programming and software engineering is arbitrary at best. Everybody understands the "engineering" in software engineering has nothing to do with other certified engineering practices, so the hair splitting here strikes me as mere gatekeeping of titles. Responsibility and accountability for outcomes have always been a requirement regardless of the title.
Probably worth clarifying with GP what responsibility and accountability they’re referring to.
Where I live, if an engineer signs off on a bridge design and the bridge personally collapses, they are personally liable for harm done to folks on the bridge. As far as I’m aware software engineering does not have something like that.
Lines of code is a poor metric for productivity, revenue, and profit. The question is: is AI making existing developer populations more efficient to where the demand trajectory for software engineers declines over time. I think we're too early to tell and that early indicators are the unsophisticated in power going "Magic robot makes you work better, use magic robot or we will replace you, and we're not going to hire materially more people until we approach what appears to be a failure mode." Bit of a collective industry death march based on vibes and cargo culting.
I agree and I think short term companies will most likely reduce their developer count because AI seems to be cheaper. However, as more code is written by AI the demand for more software will increase therefore the demand for developers will increase.
Of course I am not entertaining the idea of AGI because at that point it will no longer matter as much but if we assume that AI will improve but still require human resourcing the demand for these resources will increase. Don't you think?
Neither. Hiring and firing is the primary game of software employment. It’s why so much corporate software sucks compared to open source alternatives under strong leadership. It’s the game most developers play instead of actually writing software.
Instead I really wish there were was more emphasis, like at my current employment, all focused on ownership, training, and retention.
The last time I asked an LLM AI to review my code, it added extra lines that weren't originally there. Then, it claimed that there were many bugs in those lines. When I pointed this out, it apologized. But I don't know how much time I wasted on this thing.
I'm a developer with over 30 years of experience. Should I ease into vibe coding?
The problem isn't the overall demand for developers. The issue is that AI will replace entry-level developers, thus eventually reducing the ranks of experienced developers. A senior developer brings much more to bear on her work than simply coding skills. Where is she going to get that experience?
That is a real problem indeed. It is the same problem though. Obviously the developers that will get hired to do the 10% need to be developers with a lot of experience but as you say where you are we going to find these people if the overall number of experienced personal has dropped?
In both cases productivity increases though but I agree it will increase the valuation of certain companies tremendously. That being said I cannot see this as being a zero-sum game. We multiple on both ends of the equation.
I can tell you anecdotally as someone who only does green field implementations which are proof of concepts/MVPs to train clients and to hopefully get a larger implementation down the road where we will need to bring in a team, before LLMs got “good enough”, I would have had to staff 1-3 junior engineers to do the grunt work while I did a lot of the higher customer facing, architecture, team lead type work, now I can do it all myself in the same amount of time.
The code that even ChatGPT generates is usually just as well structured using good software practices as I would have done - and I’ve been in the field professionally since 1996.
The only work I do manually is moving functions out into their own libraries and telling it to use my libraries to reduce duplicate code.
It’s going to hurt junior devs the most. I still have to do the higher level design work and tease out requirements.
There’s almost nothing to be gained by writing about AI (particularly anything remotely positive). The anti-AI people are absolutely foaming at the mouth about how bad it is. Best to let them have their fun while the rest of us are quietly getting value out of it.
No one is going to hire more developers, that's the entire point of AI. You'll do more work and provide more value to shareholders for the same if not less pay until you're rich enough to escape the hell-cycle of capitalism or you die.
Assuming that the demand for software remains at the same level as before the infliction point. However, this is not likely to be the case (https://en.wikipedia.org/wiki/Jevons_paradox).
What I am speculating that we will see is an increasing demand for developers. While AI will be churning more code the work that cannot be done by AI will require more developers overall then ever before.
I'm not certain Jevon's Paradox really applies here. We're not talking about mining coal or cutting down trees to meet increased demand - we're talking about a future in which software is mostly written by other software, and AI's capabilities will be improving faster than human developers can be educated. The scale of software demand may increase but more of that demand will be met by AI than by human beings.
So if the demand for developers increases at all, that increase will be minimal. And bear in mind that every aspect of every software process will be optimized as much as possible for AI automation, including languages used, so the set of work that "cannot be done by AI" will also be minimized.
Peak coal jobs was in the 1980s but peak output was in the early 2000s, while employment went from 200k to about 75k in that period. Sure, coal might have been cheaper but it didn't translate into more jobs overall.
There's a divide between Engineering a system, and implementing it in most other fields. We're now facing the situation where LLMs have artificially accellerated the trend we were already on, towards a professionalization of the art of Software Engineering. The title inflation of most of the programing jobs being incorrectly labeled as "software engineer" has obscured this trend.
The demand for actual Engineering skills won't go away anytime soon. The thing about Engineers is accountability and responsibility. If you're responsible for outcomes, you must exercise authority over the process. This gives you bargaining power for wages, which FAANG and others don't want you to have.
The delineation between programming and software engineering is arbitrary at best. Everybody understands the "engineering" in software engineering has nothing to do with other certified engineering practices, so the hair splitting here strikes me as mere gatekeeping of titles. Responsibility and accountability for outcomes have always been a requirement regardless of the title.
Probably worth clarifying with GP what responsibility and accountability they’re referring to.
Where I live, if an engineer signs off on a bridge design and the bridge personally collapses, they are personally liable for harm done to folks on the bridge. As far as I’m aware software engineering does not have something like that.
Lines of code is a poor metric for productivity, revenue, and profit. The question is: is AI making existing developer populations more efficient to where the demand trajectory for software engineers declines over time. I think we're too early to tell and that early indicators are the unsophisticated in power going "Magic robot makes you work better, use magic robot or we will replace you, and we're not going to hire materially more people until we approach what appears to be a failure mode." Bit of a collective industry death march based on vibes and cargo culting.
I agree and I think short term companies will most likely reduce their developer count because AI seems to be cheaper. However, as more code is written by AI the demand for more software will increase therefore the demand for developers will increase.
Of course I am not entertaining the idea of AGI because at that point it will no longer matter as much but if we assume that AI will improve but still require human resourcing the demand for these resources will increase. Don't you think?
This is what makes sense to me at least.
Neither. Hiring and firing is the primary game of software employment. It’s why so much corporate software sucks compared to open source alternatives under strong leadership. It’s the game most developers play instead of actually writing software.
Instead I really wish there were was more emphasis, like at my current employment, all focused on ownership, training, and retention.
The last time I asked an LLM AI to review my code, it added extra lines that weren't originally there. Then, it claimed that there were many bugs in those lines. When I pointed this out, it apologized. But I don't know how much time I wasted on this thing. I'm a developer with over 30 years of experience. Should I ease into vibe coding?
The problem isn't the overall demand for developers. The issue is that AI will replace entry-level developers, thus eventually reducing the ranks of experienced developers. A senior developer brings much more to bear on her work than simply coding skills. Where is she going to get that experience?
That is a real problem indeed. It is the same problem though. Obviously the developers that will get hired to do the 10% need to be developers with a lot of experience but as you say where you are we going to find these people if the overall number of experienced personal has dropped?
BTW, I also posted a longer version of this realisation here for what's worth: https://go.cbk.ai/more-work
The demand will be for developers to be phenomenal debuggers of massive ai code bases.
Time to retire. Unless one enjoys debugging instead of coding.
As if most code bases written by humans are works of art.
AI software developers are similar to self driving cars. Both are great for increasing stock prices, but not much more than that.
In both cases productivity increases though but I agree it will increase the valuation of certain companies tremendously. That being said I cannot see this as being a zero-sum game. We multiple on both ends of the equation.
I can tell you anecdotally as someone who only does green field implementations which are proof of concepts/MVPs to train clients and to hopefully get a larger implementation down the road where we will need to bring in a team, before LLMs got “good enough”, I would have had to staff 1-3 junior engineers to do the grunt work while I did a lot of the higher customer facing, architecture, team lead type work, now I can do it all myself in the same amount of time.
The code that even ChatGPT generates is usually just as well structured using good software practices as I would have done - and I’ve been in the field professionally since 1996.
The only work I do manually is moving functions out into their own libraries and telling it to use my libraries to reduce duplicate code.
It’s going to hurt junior devs the most. I still have to do the higher level design work and tease out requirements.
There’s almost nothing to be gained by writing about AI (particularly anything remotely positive). The anti-AI people are absolutely foaming at the mouth about how bad it is. Best to let them have their fun while the rest of us are quietly getting value out of it.
No one is going to hire more developers, that's the entire point of AI. You'll do more work and provide more value to shareholders for the same if not less pay until you're rich enough to escape the hell-cycle of capitalism or you die.
Assuming that the demand for software remains at the same level as before the infliction point. However, this is not likely to be the case (https://en.wikipedia.org/wiki/Jevons_paradox).
What I am speculating that we will see is an increasing demand for developers. While AI will be churning more code the work that cannot be done by AI will require more developers overall then ever before.
I'm not certain Jevon's Paradox really applies here. We're not talking about mining coal or cutting down trees to meet increased demand - we're talking about a future in which software is mostly written by other software, and AI's capabilities will be improving faster than human developers can be educated. The scale of software demand may increase but more of that demand will be met by AI than by human beings.
So if the demand for developers increases at all, that increase will be minimal. And bear in mind that every aspect of every software process will be optimized as much as possible for AI automation, including languages used, so the set of work that "cannot be done by AI" will also be minimized.
Even in the case of coal mining, jobs fell as output increased. See the two charts in https://acoel.org/the-rise-and-decline-of-coal-mining-jobs-1...
Peak coal jobs was in the 1980s but peak output was in the early 2000s, while employment went from 200k to about 75k in that period. Sure, coal might have been cheaper but it didn't translate into more jobs overall.