I retired from paid sw dev work in 2020 when COVID arrived.
I’ve worked on my small projects since with all development by hand. I’d followed the rise of AI, but not used it.
Late last year I started a project that included reverse engineering some firmware that runs on an Intel 8096 based embedded processor. I’d never worked on that processor before. There are tools available, but they cost many $. So, I started to think about a simple disassembler.
2 weeks ago we decided to try Claude to see what it could do. We now have a disassembler, assembler and a partially working emulator. No doubt there are bugs and missing features and the code is a bit messy, but boy has it sped up the work.
One thing did occur to me. Vendors of small utilities could be in trouble. For example I needed to cut out some pages from a pdf. I could have found a tool online(I’m sure there are several), write one myself. However, Claude quickly performed the task.
This is a mix of the “in the future, everyone will have a 3D printer at home and just 3D print random parts they need” and “anyone can trivially build Dropbox with rsync themselves” arguments.
Tech savvy users who know how to use LLMs aren’t how vendors of small utilities stay in business.
They stay in business because they sell things to users who are truly clueless with tech (99% of the population, which can’t even figure out the settings app on their phone), and solid distribution/marketing is how you reach those users and can’t really be trivially hacked because everyone is trying to hack it.
Or they stay in business because they offer some sort of guarantee (whether legal, technical, or other) that the users don’t want to burden themselves with because they have other, more important stuff to worry about.
I don't know. It's one thing to tell Joe or Jane User to "Get an FTP account, mount it locally with curlftpfs, and then use SVN or CVS on the mounted filesystem." But if Joe or Jane can just cut-and-paste that advice into a prompt and get their own personal Dropbox...
Except when that new Dropbox fails Joe or Jane on that Saturday evening, their only recourse is to ask the AI for help, and the AI starts spinning “oh yeah, mmm, I think I found where the problem is. Cut and paste these debugging lines in that function and let me know what the output is…”
Meanwhile, this year, that happens less often than it did last year... and it actually isn't how AI-assisted development works at all. Agentic models do the cutting-and-pasting by themselves, evaluate the results by themselves, and almost always succeed at fixing the problem by themselves.
> Vendors of small utilities could be in trouble. For example I needed to cut out some pages from a pdf. I could have found a tool online(I’m sure there are several), write one myself. However, Claude quickly performed the task.
Definitely. Making small, single-purpose utilities with LLMs is almost as easy these days as googling for them on-line - much easier, in fact, if you account for time spent filtering out all the malware, adware, "to finish the process, register an account" and plain broken "tools" that dominate SERP.
Case in point, last time my wife needed to generate a few QR codes for some printouts for an NGO event, I just had LLM make one as a static, single-page client-side tool and hosted it myself -- because that was the fastest way to guarantee it's fast, reliable, free of surveillance economy bullshit, and doesn't employ URL shorteners (surprisingly common pattern that sometimes becomes a nasty problem down the line; see e.g. a high-profile case of some QR codes on food products leading to porn sites after shortlink got recycled).
Having an LLM spit out a few hundred lines of HTML and JavaScript is not a colossal waste of resources, it's equivalent to running a microwave for a couple of seconds.
Not to mention, my little tool is using much less electricity running than just about anything else I could easily find on-line, simply by the virtue of being minimal, and completely free of superfluous visual bullshit, upsells, tracking, telemetry, and other such secondary aspects of anything people publish and advertise for others to use.
Don't get the anti-AI propaganda get to you too much. Inference is cheap on the margin.
Consider: there are models capable (if barely) of doing this job, that you can run locally, on a upper-mid-range PC with high-end consumer GPU. Take that as a baseline, assume it takes a day instead of an hour because of inference speed, tally up total electricity cost. It's not much. Won't boil oceans any more than people playing AAA video games all day will.
Sure, the big LLMs from SOTA vendors use more GPUs/TPUs for inference, but this means they finish much faster. Plus, commercial vendors have lots of optimizations (batch processing, large caches, etc.), and data centers are much more power-efficient than your local machine, so "how much it'd cost me in power bill if I did it locally" is a good starting estimate.
It's not even "users", just the user. Nice thing about LLMs is that it's cheap to develop small tools tailor-made for audience of few, or in this case, just one.
1) This was for my wife. She is not proficient in Linux or CLI in general, and (like ~all white collar workers these days) works almost exclusively in browser tools (exception being pre-O365 versions of Word and Excel we keep running on her laptop because she prefers them).
2) I never heard of `qrencode` CLI tool until today. For some reason I didn't even consider it might exist (maybe because last time I checked, which was many years ago, there was none).
4) Even if I knew about it, I'd still have to build a web frontend for it, and I'd need a proper server for it, which I'd then have to maintain properly, and secure it against the `qrencode` call becoming an attack vector.
So frankly, for my specific problem, my solution is strictly better.
I briefly worked in a team that implemented a JVM on a mobile OS (before the iPhone) and one of the senior devs said Jazelle was in effect very inefficient because of all the context switching between ARM mode and Jazelle mode.
Turned out a carefully tuned ARM JVM was in practice th best
I’ve tried 2 AI tools recently. Neither could produce the correct code to calculate the CPU temperature on a Raspberry Pi RP2040. The code worked, looked ok and even produced reasonable looking results - until I put a finger on the chip and thus raised the temp. The calculated temperature went down.
As an aside the free version of chatGPT didn’t know about anything newer than 2023 so couldn’t tell me about the RP2350
How can you be sure putting the finger on the chip raise the temp? If you feel hot that means heat from the chip is being transferred to your finger, that may decrease the temp, no?
wouldn't your finger have acted as a heat sink, lowering the temp? sounds like the program may have worked correctly. could be worth trying again with a hot enough piece of metal instead of your finger
Ha. I asked it to write some code for the Raspberry Pi RP2350. It told me there might be some confusion as there is no official product release of the RP2350.
If it doesn’t know that, then what else doesn’t it know?
Scarily close to satire of humans in denial about AI capabilities (not saying that it's the case here but I can imagine easily such arguments when AI is almost everywhere superhuman)
I just checked. The code it gave me, though syntactically correct, was wrong functionally. The rp2040 temp reading increases and the ADC value decreases. ChatGPT didn’t invert the values.
It's doable, but it took me closer to 20 years.
I got to zero net assets in July 2001. Retired from paid work (mainly sw eng contracting) at the start of covid in April 2020.
I should say, at the start I wasn't married and had no dependents. Also, for large parts of those 20 years, I didn't need to own or use a car.
I tried code gen for the first time recently. The generated code look great, was commented and ran perfectly. The results were completely wrong.
The code was to calculate the cpu temperature from the Raspberry Pi RP2350 in python.
The initial value look about right, then I put my finger on the chip and the temp went down!
I assume the model had been trained on broken code. This lead me to think how do they validate code does what it says
Nobody is saying that you don't have to read and check the code. Especially for things like numerical constants. Those are very frequently hallucinated (unless it's something super common like pi).
I’ve now retired from professional programming and I’m now in hobby mode. I learn nothing from reading AI generated code. I might as well read the stack overflow questions myself and learn.
Did you review the code itself, or test the code beyond just putting your finger on the chip? Is it possible that your finger was actually cooler than the chip and acted as a heat sink upon contact?
The code looked fine. And I don’t think my finger is colder than the chip - I’m not the iceman.
The error is the analog value read by the ADC gets lower as the temperature rises.
30 odd years ago, part of my role was to colour balance cameras in a studio environment. We didn’t need computers - but at most there were only 5 cameras :)
I think Newton is on the phone...
Once the accelerating force is removed the object will go into an orbit that includes the point at which the force stops. This however is Earth
It doesn't remove fuel entirely, they still need an engine to circularize their orbit.
> The velocity boost provided by the accelerator's electric drive results in a 4x reduction in the fuel required to reach orbit, a 10x reduction in cost, and the ability to launch multiple times per day.
Wow are there issues with making things like chips that can even survive that? I remember when I was a scientist and 10,000g was some very high speed centrifuging that you better make sure everything like even glass and plastic can handle.
Unless it reaches escape velocity (ie its never coming back) then it will always comeback to where it started from. The object is under the same gravitaional influences as the Earth
As others have pointed out, there is a rocket motor to put the payload into an orbit.
reply