Why aren’t they dogfooding their own products to replace such roles?
And seeing how people use it: good programmers review output and iterate to get better output. But bad programmers simply trust the output is good: they have no ability to review it themselves and often don't try.
With about 5-10h over the weekend using free tier Claude and ChatGPT I managed to put together a scraper for a particular thing on a website I’m interested, grab the item images, do an initial pass with local OCR, if it hit some keywords, run openCV to crop for better OCR and dump the hits for further investigation.
Nothing particularly advanced but it would have taken me a horrendous amount of time without it to be half as good, like it did when I built a similar scraper 10 years ago.
Neither were very good code quality i’m sure.
An expert can either use the tool more effectively, or see all the issues in a less experienced person's output.
Both of these are good things, the mistake a ton of people are making is experiencing industrial scale Dunning-Kruger and thinking "Only my expertise is still valuable, every other white collar role is done!"
The second-order mistake is thinking that raising the floor like that devalues expertise instead of increasing demand for it. The net-effect of me starting to play with CAD because it's a little easier now isn't that I don't hire my friends who are experts to make a tiny spacer I'm going to 3d print, I never would have hired them for that, it's that maybe I start learning the skills and decide to take on a more ambitious project where I do need to hire one of them for some help, or start ordering custom CNC'd parts -- scale that to the entire economy.
I guess this means that you have some good instincts or habits that would be good for a programmer, even if you didn't choose that path.
Programming is more than just knowing the syntax of the programming language and the APIs you want to use. It also requires clear communication and clear thinking, checking things, etc.
There is no reason why a non-programmer couldn't also think clearly or carefully. It's just a fact about humanity that most people don't; and many people have jobs that do not require this, so they never develop the skills. Some people develop them for job-unrelated reasons.
Now we are at the moment when the LLMs can do the syntax and APIs for you, but they still fail at clear thinking and proper caution. That elevates a good potential programmer to a good vibecoder.
On average, the output is still better than what a bad programmer would produce.
If I find a task to be annoying and repetetive, I use an LLM to do it, then validate it. The quality of my work only improves, and so does the speed, and I don't lose my skill over time.
Once I got the manpages ready and the way I wanted, I just stored those in an empty project folder and told the agent to implement as specified in a language of my choosing.
Maybe for other development scenarios like GUI there is some "native" way to spec the project in a common format like this.
For example, a recent story about the openclaw creator using $1.3M of tokens/month. And let's assume he's getting paid $5M/yr which is probably a vast under estimate.
Is he providing value that a traditional software development org with normal developers couldn't provide for $20M/yr?
Yes, $1.3M in token cost in less than 30 days and some days were even off-peak, if you can call it that with that insane scale that likely hides quite a lot of tokens in the lower bars.
Finally in some ways agentic workflows magnify the power of the individual who is adept at harnessing them, they don’t have to argue (much) with the agents to effect their ideas. I’ve found a lot of very bright engineers spend their days fighting to be heard by managers and peers who can’t / won’t understand them. By unshackling them from trying to debate down idiots, they deliver way way more, and of the right things, than they otherwise could have.
Leibniz, literally parallel.
Was Newton just a smart guy at the right place and the right time. These smart folks require other smart folks to understand and verify what they did. There are many who have amazing pedigrees in history.
I don't think existing companies will bite that bullet, but I think you'll see AI native companies in five years with like, a baffling small number of people.
Getting bought by the bigger companies as it is currently the goal of most start-up since no one cares about monopolies and anti competitive behavior nowadays.
Who claimed that?
Their customers will be happy if their product replaces all the junior positions and midwit developers off the payroll. then that's already a huge saving to any company's bottom line.
Even if it doesn't directly replace workers, reducing the bargaining power of those spoiled SW devs and not having to give them huge raises all the time or they leave, is still enough. That's the whole point of layoffs and offshoring anyway.
Possibly not if they are paying the full cost of inference
Dario Amodei
Oh, they do. That's why they buy actual engineers to fix the vibe-coded slop they produce en masse: https://x.com/jarredsumner/status/2026497606575398987
Unfortunately those engineers slowly succumb to the same slopcoding vibes, so they go and buy a bunch of new engineers.