Coding boot camps fight for relevance in the era of generative AI and a shrinking job market

midian182

Posts: 10,235   +138
Staff member
A hot potato: Tech leaders such as Tim Cook spent years telling people that if they wanted a lucrative career in the industry, they needed to learn to code. But the advent of generative AI now means that coding isn't the guarantee of a good job that it once was. The number of active job postings for software developers has dropped 56%, and the number of placements on coding boot camp courses is dwindling.

In 2024 alone, 519 companies in the tech industry have laid off just over 149,000 people. A substantial percentage of those layoffs come from firms "streamlining" their businesses by using generative AI to do a human's job or assist with certain tasks.

The New York Times reports that ChatGPT and similar tools are also impacting coding boot camps. In 2020, a survey of 3,000 graduates found 79% said the courses had helped them get a job in tech, along with an average salary increase of 56%. But the outlook for graduates has changed drastically over the last few years.

One recent coding boot camp graduate told The NYT they had yet to land a single interview since completing their course.

The publication writes that a founder of one of the camps decided to pause his courses indefinitely this year because his job placement rates had fallen from 90% to below 60%.

The grimmest statistic comes from data compiled by CompTIA: active job postings for software developers have dropped 56%, while postings for inexperienced developers are down 67%.

"I would say this is the worst environment for entry-level jobs in tech, period, that I've seen in 25 years," said Venky Ganesan, a partner at the venture capital firm Menlo Ventures.

Each new iteration of generative AI tools tends to be better at writing code than the last. In September, OpenAI unveiled a new series of AI language models named the "o1," specifically engineered to enhance reasoning capabilities, particularly for complex issues in science, coding, and mathematics.

Some experts say AI tools still lack the essential coding skills that even junior programmers possess, and that LLMs like GPT-4 continue to make stupid mistakes and are incapable of performing some tasks well, if at all.

Advocates, on the other hand, say AI is good for programming, helping speed up the process. About 60% of 65,000 developers surveyed in May by StackOverflow say they had used AI coding tools this year. The assistants may not have boosted their productivity or effectively helped prevent burnout, though, according to a recent study.

Probably the biggest seal of approval AI received came last month when Google CEO Sundar Pichai revealed that 25% of new code the company creates is written by the technology.

There's no hiding from the fact that generative AI has changed the industry, from the many job losses to the computer science coding classes that now deemphasize syntax. Some coding camps are turning into AI and machine learning boot camps, and the phrase "AI monkey" has made its way into common parlance.

Jensen Huang, CEO of the company whose hardware is driving the AI revolution, painted the clearest picture of this industry shift at the start of the year. He said that AI has made the need to learn programming redundant, and humans should instead focus on more important skills such as biology, education, manufacturing, or farming.

Permalink to story:

 
Speaking as a Software Engineer with 15 years of practical experience:

AI is "great" at fulfilling simple, well defined tasks. The problem is once the size of the problem reaches a point where you can no longer describe it in a sentence or two, AI models start to break down and generate either useless code (best case) or code that does something other then what you want (worse).

Thing is, the above also generally applies to many of the individuals who come out of these camps. Because they more often then not don't learn how to "program", they learn a syntax and how to solve really simple (but more often then not, useless) problems with that syntax.

Maybe I'm biased; I work primarily with embedded systems where I'm generally only a level or two away from direct hardware access. But the one thing I've seen repeatedly, from both AI and inexperienced Software Engineers, is the inability to define, let alone solve, the requirements for a given piece of software that has to be created.

Basically: Learn how to design software, rather then a specific syntax. Learn best practices when it comes to threading and memory management. Learn how to debug and isolate problems that will inevitably occur. Do this and you'll be irreplaceable.
 
Speaking as a Software Engineer with 15 years of practical experience:

AI is "great" at fulfilling simple, well defined tasks. The problem is once the size of the problem reaches a point where you can no longer describe it in a sentence or two, AI models start to break down and generate either useless code (best case) or code that does something other then what you want (worse).

Thing is, the above also generally applies to many of the individuals who come out of these camps. Because they more often then not don't learn how to "program", they learn a syntax and how to solve really simple (but more often then not, useless) problems with that syntax.

Maybe I'm biased; I work primarily with embedded systems where I'm generally only a level or two away from direct hardware access. But the one thing I've seen repeatedly, from both AI and inexperienced Software Engineers, is the inability to define, let alone solve, the requirements for a given piece of software that has to be created.

Basically: Learn how to design software, rather then a specific syntax. Learn best practices when it comes to threading and memory management. Learn how to debug and isolate problems that will inevitably occur. Do this and you'll be irreplaceable.
The main issue is that coding camps do not produce software engineers, but coders. For a while companies were forced (by the scarcity of skilled labor) to hire coders and try to make them productive. The results were not always the best. Remains to be seen if AI/LLM or whatever comes next can indeed make software engineers obsolete. I hope they do because I'm a bit tired but at the same time I can't let go :)
 
The main issue is that coding camps do not produce software engineers, but coders. For a while companies were forced (by the scarcity of skilled labor) to hire coders and try to make them productive. The results were not always the best. Remains to be seen if AI/LLM or whatever comes next can indeed make software engineers obsolete. I hope they do because I'm a bit tired but at the same time I can't let go :)
Personally, I don't see it. My own work is trying to offload our Engineering to AI/LLMs, and it is *not* going well. The problem is the impossibility of describing the problem, the requirements, the lessons learned from prior similar work, and getting the AI to come up with something that makes any amount of logical sense.

Where AI is going to get a foothold is making low-level Software Engineers (<5 years experience) obsolete, since that's about where it's at. And that is going to cause some long-term problems in the industry. But at least in the short term, people like me will make a killing fixing up whatever poorly designed software gets put into production.
 
This is weird, like a self destructive solution.

Those experienced software guys were the inexperienced guys once, and by cutting their teeth on complex problems they became the experienced guys.

Tech companies better have a development program in place for AI monkeys to actually learn the code their AI is generating so that as the senior people age out coding groups don't suffer a knowledge/doctrine collapse because all anyone knows is how to ask the AI for code, not how to verify that the code is actually any good.
 
This is weird, like a self destructive solution.

Those experienced software guys were the inexperienced guys once, and by cutting their teeth on complex problems they became the experienced guys.

Tech companies better have a development program in place for AI monkeys to actually learn the code their AI is generating so that as the senior people age out coding groups don't suffer a knowledge/doctrine collapse because all anyone knows is how to ask the AI for code, not how to verify that the code is actually any good.

Hahahahhahaha no.

I work where we have to support codebases on the timescale of decades. 90% of our SE department is on the wrong side of 60. And there's *no* knowledge retention program. For the simple reason that it's an expense that doesn't result in immediate revenue.
 
This is weird, like a self destructive solution.

Those experienced software guys were the inexperienced guys once, and by cutting their teeth on complex problems they became the experienced guys.

Tech companies better have a development program in place for AI monkeys to actually learn the code their AI is generating so that as the senior people age out coding groups don't suffer a knowledge/doctrine collapse because all anyone knows is how to ask the AI for code, not how to verify that the code is actually any good.
The software industry is heading for the same iceberg that skilled trades and skilled manufacturing labor hit decades ago. The transition to younger workers is gonna be PAINFUL. Ford found this out the hardw ay, they laid off all their old die makers in 2008, then figured out they had almost noone young who could do it. So they hired them back. Still refused to train young guys. Now the die makers are aging out again, and are not coming back a 2nd time. So now Ford has to outsource to third parties made up of the few young guys who get this stuff, who are charging them by the kidney for the work. Same for machining shops, basic work has gone from a few days turn-around to months, with bigger jobs flat out no longer accepted, as the old guys retire and nobody replaces them.

Welcome to the competency crisis. Get comfy and buy a bucket of popcorn while you can, its gonna get WAY worse over the next 20 years as people figure out you cant eat social media posts or live in AI generated houses. Plus side, if you're younger and have a skill, you are gonna make BANK.
 
Not gonna lie, this AI thing is handy at times. I needed to write some complex Excel formulas and Copilot in Bing was very precise in its answers, including copy and paste functionality. I can image the routine coding it can do. ROUTINE, mind you. Creativity is still for now a human trait.
 
The software industry is heading for the same iceberg that skilled trades and skilled manufacturing labor hit decades ago. The transition to younger workers is gonna be PAINFUL. Ford found this out the hardw ay, they laid off all their old die makers in 2008, then figured out they had almost noone young who could do it. So they hired them back. Still refused to train young guys. Now the die makers are aging out again, and are not coming back a 2nd time. So now Ford has to outsource to third parties made up of the few young guys who get this stuff, who are charging them by the kidney for the work. Same for machining shops, basic work has gone from a few days turn-around to months, with bigger jobs flat out no longer accepted, as the old guys retire and nobody replaces them.

Welcome to the competency crisis. Get comfy and buy a bucket of popcorn while you can, its gonna get WAY worse over the next 20 years as people figure out you cant eat social media posts or live in AI generated houses. Plus side, if you're younger and have a skill, you are gonna make BANK.

Exactly this.

Companies don't want to invest anymore; investments don't show up as positive revenue. Our entire financial system is built around the next quarterly report, which is why you see so many companies across a variety of different industries running into the same exact problems; the issues facing a company like Boeing aren't really any different then the issues affecting EA: too many layers of management, lack of internal investment, and making financial decisions rather then engineering ones.

Pretty much everywhere I've worked, I see companies held together by one or two boomers on the wrong side of retirement age, with no knowledge retention or training of any kind. Everyone sees the writing on the wall, and knows how bad things will be in just a handful of years (and a bunch of us are ready to bail the second things go downhill; always have a backup plan after all).
 
Back