Earlier quoted context omitted.
> However, this part of AI evolves very quickly. So given these are known problems, why shouldn't we expect rapid improvements in agentic AI systems for software development, to the point where software developers who stick with the old paradigm will indeed be eroded in time? Because writing code has always been the easy part. A senior isn't someone who's better at writing code than a junior - they might well be wors…
I don’t know what level of experience you have with agentic AI, but the frontier models are also really good at things like product management and data modeling. You can start with a description of the problem and end up with a really solid design plan that you can then give the AI to implement. So yeah, if you’re starting with a “write me code that does X Y Z” then you aren’t getting the most out of these tools, bec…
Software development in the time of strange new angels
51–60 of 71 posts
Re: Software development in the time of strange new angels
#52Earlier quoted context omitted.
> In 2022 the best available models was GPT-3 text-davinci-003 at $60/million input tokens. >GPT-5 today is $1.25/million input tokens - 48x cheaper for a massively more capable model. Yes - but. GPT-5 and all the other modern "reasoning models" and tools burn through way more tokens to answer the same prompts. As you said: > We're beginning to find more expensive ways to use the models though. Coding Agents like Cla…
That really is the most interesting question for me: when will it be possible to run a model that is good enough to drive Claude Code or Codex CLI on consumer hardware? gpt-oss-120b fits on a $4000 NVIDIA Spark and can be used by Codex - it's OK but still nowhere near the bigger ones: https://til.simonwillison.net/llms/codex-spark-gpt-oss But... MiniMax M2 benchmarks close to Sonnet 4 and is 230B - too big for one Sp…
Trouble is, there's not even much hype surrounding the launch yet, much less shipping hardware. Which seems kind of ominous.
Re: Software development in the time of strange new angels
#53The core of the entire argument is that the $150/hour is based on a developers ability to physically write code, which is not true. Having something that can generate code reliabily (which these things can barely do even with an expert at the wheel) doesn't address any of the actual hard problems we deal with on a daily basis. Plus running AI tools is going to get much more expensive. The current prices aren't sustai…
Open source AI is getting cheaper and cheaper. Model companies run inference at a profit, the lack of profitability from AI companies is just due to them putting all their capital into training the next generation of models.
Re: Software development in the time of strange new angels
#54Earlier quoted context omitted.
> skill of an all-around businessman So do you imagine that AI will reach the point that a business guy will say make me a web site to blah blah blah, and the AI will say sure boss and it will appear? Sort of what a dev/team of devs/testers/product managers/BA's would do now? the current batch is a long way from this afaik
Maybe. I have a fair amount of uncertainty about the speed of AI development but I think that that is well within the realm of possibility (and definitely a possibility developers should be considering). Note that even if AI replaces an entire dev team, it's still not apparent that the process would be as easy as you're saying (at least for a while after that), since after all even with a crack dev team products are…
The other thing I've been thinking is that most corporates now are mainly software (so it's been said), if thats the case and software becomes cheaper the barrier to entry to compete with corporates lowers, they become ripe for disruption. Insurance comes to mind, same with banking, probably others, search? its already disrupted, new industries will probably arise to - data validation, for example, is going to be an issue in the age of AI. The idea that making web sites for a living for the next century was probably always a very wishful way of thinking, but the idea that software development will disappear is also naive imho. However it's yet to be proved that software dev is cheaper with AI.
Re: Software development in the time of strange new angels
#55The core of the entire argument is that the $150/hour is based on a developers ability to physically write code, which is not true. Having something that can generate code reliabily (which these things can barely do even with an expert at the wheel) doesn't address any of the actual hard problems we deal with on a daily basis. Plus running AI tools is going to get much more expensive. The current prices aren't sustai…
Why would the ai model makers charge less than $149/hr ?
Why hasn't outsourcing attacked equal chunks of that $150/hr all these years now?
If companies dont realize that if an employee is required 1 hour per week, they still need a full salary that covers all the rising costs of housing/food/health/necessity... then most knowledge workers just die. Even more so in up and coming countries, its just massive world war and suffering if we dont change capitalism somehow. Why would AI employees keep working toward such destruction and death?
What if the 80/20 problem is more reality, machine learning and LLMs can never get that last 20 percent working right, but now nobody knows how to finish the last chunks? Seems more like the last 10% of coders that dont die, should be charging hundreds of times more $/hr. Even this doesnt fix the problem because the knowledgeable will die off and nobody hired juniors for years.
You think housing prices can sustain their value in this nonsense? Old people who are still in charge of everything and own everything will destroy all of this before it starts affecting their assets
Re: Software development in the time of strange new angels
#56The core of the entire argument is that the $150/hour is based on a developers ability to physically write code, which is not true. Having something that can generate code reliabily (which these things can barely do even with an expert at the wheel) doesn't address any of the actual hard problems we deal with on a daily basis. Plus running AI tools is going to get much more expensive. The current prices aren't sustai…
Completely agree on your first point: software development is so much more than writing code. LLMs are a threat to programmers for whom the job is 8 hours a day of writing code to detailed specifications provided by other people. I can't remember any point in my own career where I worked with people who got to do that. There's a great example of that in the linked post itself: > Let's build a property-based testing s…
Re: Software development in the time of strange new angels
#57> You might be expecting that here is where I would start proclaiming the death of software development.... I'm not going to do that, because I absolutely don't believe it. Agentic AI means that anything you know [how] to code can be coded very rapidly. Read that sentence carefully. If you know just what code needs to be created to solve an issue you want, the angels will grant you that code at the cost of a prompt o…
> skill of an all-around businessman So do you imagine that AI will reach the point that a business guy will say make me a web site to blah blah blah, and the AI will say sure boss and it will appear? Sort of what a dev/team of devs/testers/product managers/BA's would do now? the current batch is a long way from this afaik
There is lots of founder developers, there is a lot of people who can configure lots of complex tech but are just not strictly speaking coders.
Re: Software development in the time of strange new angels
#58It is encumbant on us as devs to use this tool and understand it.
The invention of the chainsaw did not eliminate the lumberjack as a profession. Lumberjacks learned how to become more productive with this dangerous new tool.
Re: Software development in the time of strange new angels
#59Earlier quoted context omitted.
Netflix was initially $7/mo in 2007. Currently it goes for $18/mo in 2025. 150% increase in 18 years is about triple the rate of inflation I'd call that more than "a little bit", but I'd agree that if LLMs go the same way it probably doesn't change the equation much at all. Plus they do still have an $8/mo ad-supported plan. After adjusting for inflation, that's actually cheaper than the original!
I think it's gonna be a lot worse with LLMs. Mainly because they're substantially under charging, I don't think the cost of operating is going to drop much, and the workflows are only going to get more token hungry. The incentives here are also fucking atrocious. They aren't incentivised to make the model as good as possible. It's in their best interest to tune so it's good enough to not drive you off, but bad enough…
It wont: if/when LLMs start to get too expensive, people will just migrate to open models, run it local, etc. I see no scenario where we are held hostage by the main providers.
Re: Software development in the time of strange new angels
#60The core of the entire argument is that the $150/hour is based on a developers ability to physically write code, which is not true. Having something that can generate code reliabily (which these things can barely do even with an expert at the wheel) doesn't address any of the actual hard problems we deal with on a daily basis. Plus running AI tools is going to get much more expensive. The current prices aren't sustai…
Completely agree on your first point: software development is so much more than writing code. LLMs are a threat to programmers for whom the job is 8 hours a day of writing code to detailed specifications provided by other people. I can't remember any point in my own career where I worked with people who got to do that. There's a great example of that in the linked post itself: > Let's build a property-based testing s…
Yeah but eventually there wont be enough people who actually do know all that.
The hope amongst the proponents is that by the time that happens they wont need anyone who knows that because SOTA will have replaced those people too.