> Things are going really well for us actually,
Not right now anymore. Have you seen the layoffs? Things went well because of years of cheap cash. The music is stopping and there will be even more layoffs before this is all done.
AI being able to code is going to be like what happened when construction workers started using power tools: now that one man can saw boards as fast as 5 others with handsaws, wouldn't we see a lot less demand for carpenters? What happened is that per-worker productivity went up because of the new tools and the construction materials and techniques they use were adapted to take advantage of that, there was still plenty of work, and we can build better things faster than before.
In software, imagine how much more effective engineers can be if the grunt work could be more automated. People loved autocomplete and smart IDE features and graphical debuggers because they made coding more productive. A lot of those things already do write some code. AI provides a much better version of that.
People whose only contribution to software is the grunt work of writing simple code might need to worry, but there is a lot more to working as a software engineer than grinding code. These AIs are also repeaters of existing solutions -- maybe they can solve simple problems that have pre-existing solutions in their training corpus, but an AI is not going to be able to create the next Redis or Kubernetes or AWS simply by recombining existing text.
Regarding layoffs:
Even after all the layoffs most companies still have more employees than they did in 2020. Many of the people being laid off today joined since then. They will get other jobs, maybe not in a few weeks like what used to be possible, but still. This is not like 2001 when entire companies were going out of business and laying off 100% of their employees.
A lot of the the American tech industry developed when interest rates were higher than they are now. What we had in the past decade was not normal, and in some ways it was pretty crazy -- investors didn't care if companies were sustainably operated, and now they do, and that's good for the long-term health of the industry.
Tech is not immune from the business cycle, because nothing really is, but this really is not as bad as previous downturns. Even if it gets worse before it gets better, the tech industry is fundamentally strong and will handle it.
New industries will have to spawn from generative AI for there to be a net positive. Most IT and tech jobs are lower level coding positions. This will all but eliminate these roles. SWEs especially top ones will likely be safe. But that doesn't solve the societal problem.
Regarding layoffs - we've barely touched the surface. We haven't even entered recession yet, and a financial crisis looms. Most are on hiring freeze now and just in the last 2 weeks both meta and amazon slashed additional jobs. More will be coming. The industry won't completely collapse because it's a mature industry now - FAANGs are now like Coke. Blue chip. This says nothing about its workforce though - there's nothing these companies would love more than to be able to reduce the cost of their software devs.
> Most IT and tech jobs are lower level coding positions.
I don't really buy that. It's true that most positions are IC positions, i.e. not managers, but quite a lot of them require domain knowledge and human intervention in edge cases which can't be handled by automation. If automation could solve all the problems, why are so many people oncall?
> I don't really buy that. It's true that most positions are IC positions, i.e. not managers, but quite a lot of them require domain knowledge and human intervention in edge cases which can't be handled by automation. If automation could solve all the problems, why are so many people oncall?
I don't mean IC vs manager. In any case, at my company our support staff has been reduced by 75%. In my previous life as a mechanical engineer, I've seen control systems replace 90% of powerplant staff. Also, generative AI just became a thing - because SWEs can be more efficient, you simply need less, UNLESS there is growth or a new market emerges (certainly possible), in any case millions will be left behind, just as what happened in the rust belt when manufacturing collapsed.
In any case, I hope you're right and I'm wrong :)
Designing large coherent systems for a business use case is hard. I don't think ChatGPT will ever manage that.
Plus let's be honest, the technical problems are the tip of the iceberg for software architecture. If an AI ever manages to convince my director to pay for things I'd be happy for the help.