Hardware has evolved in super interesting ways since the 90s. Multiple cores, fast drives, tons of memory, incredible vectorization, advanced instruction optimization, larger caches, fast networking, VR/AR, pocket sized computers, touch screens, etc. Pretty cool stuff. And for high performance some of that mechanical sympathy is still needed.
I don't know if I agree with that take though, because AI is, at the end of the day, an algorithm. And it's super cool from that angle, and what it can do is kind of amazing, that a computer is doing that is pretty awesome computer stuff. But I agree that wanting to only use AI to code seems like you're just in it for the results and not the process.
> I don't know if I agree with that take though, because AI is, at the end of the day, an algorithm Yes and it helps to consider AI as Algorithmic Imitation. This abbreviation carries much more precision and does not obfuscate natural intelligence which is so much more than just logic and inference.
I hate computers. The physical box things that can only break. Faulty ram, managing thermal, breaking fans all those stupid things. I love their computational model tho.
I don’t get it, at ALL. Don’t you LIKE understanding the magic of the machine?I guess that is because she has a very romantic view of a programmer's life, and also probably because she works mostly on security and reverse engineering which is pretty fun. But programming is about a bunch of different things, some of them fun, some of them not so, and it all depends on the context (person, task, deadlines, etc) Sometimes it is fun to learn the intricacies of JavaFX or whatever, and other times you just want the UI you have in mind and move forward. Sometimes it is great to get to look at the assembly and move instructions around to make it faster, some other times it is ok to make a thing that is enough but move faster, that is why we created high level languages.
Sadly this view is in direct opposition to business' interests. We have to remember that we're employees after all, and the "artisan" mindset (which I share) is not applicable to our jobs. For personal projects though, I agree.
> For personal projects though, I agree.
But even for personal projects, artisan/handmade and understanding are in a spectrum. We use libraries and compilers for example, which in a way are delegating this "magic of the machine" to something else. Sometimes I don't want to understand how md5-hash works, but just want to use it, even if it is super interesting.
@jpmonettas I think it depends, if there is a layer above that is interesting or not. When you enter into the natural language, and probabilistic probing layer as your control layer, it does change quite a bit in nature, from what was using a computational language. A high-level computer language doesn't change in nature like that.
@didibus my argument was just against that ... I don't get it, at ALL. Don't you *LIKE* understanding the magic of the machine? ... and then inferring programmers don't like computers.
also, pardon my grandiose, code becomes less and less interesting the more you understand
Ya, but I think that's not a good argument, because you're trading say the details of md5-hash, for the details of your data model or the structure of your rendering loop. You're just shifting your interest around but it remains focused on computational modeling. When you switch to agentic coding, you're now moving away from a computer language and into a natural language. It doesn't scratch the same itch.
> When you switch to agentic coding, you're now moving away from a computer language and into a natural language I don't think that is necessarily true. "Agentic coding" could mean natural language + python + pseudo code + diagrams + math + a video of me handling the mock of the device I'm trying to implement.
To make a possibly poor analogy, I feel it's like you used to be the Formula 1 driver, and now you switched to being the Race engineer (the person the driver talks to on their headset during the race). Some people might enjoy that, even more so than driving the car, but it's definitely not the same feel, even if it results in somewhat the same thing in helping win the race.
But that is an all or nothing argument, where you just get to direct the thing using natural language (or some more efficient encoding for F1 drivers). But I don't think it is the same
> Agentic coding" could mean natural language + python + pseudo code + diagrams + math + a video of me handling the mock of the device I'm trying to implement. It could, but it more and more means a zero-code/math workflow. If you want to keep up, you'll be operating more in a loop that follows: -> have AI write the design specs, review it using natural comments, have AI implement the code, review it using natural comments. Can ask the AI to produce diagrams to help you review.
If you purposely use-AI sparingly, than I'm assuming you are not the target of OPs quote.
I mean, you can choose to just use natural language if that is enough for the thing you need, but you can be as specific as you want I guess. Probably even more than before, because agent can even recommend things to look into that you didn't even know about before.
> If you purposely use-AI sparingly, than I'm assuming you are not the target of OPs quote. I was just trying to give an answer to her frustration, and that inference that "I’m convinced that a large % of programmers don’t actually like computers ..." because "... Don’t you LIKE understanding the magic of the machine?". And I guess that frustration comes from looking at a small part of what programming is about. I've heard similar arguments many times from the https://handmade.network/. Like if you don't understand what is going infinitely deep under the hood you don't know what you are doing, you are just plumbing other peoples programs or something.
Right, I think my reading was different, and more about the fact that a number of people immediately jumped shipped to just, I tell the computer in plain English what I want and it gives it to me. As-if, they were never interested in the programming at all, and it was always just an annoying and boring means to an end. Which is different than the argument that if you're not interested in every aspect or at all depth of the stack maybe you don't like programming.
Yeah, but that argument is the same you can make (and I probably heard from the https://handmade.network/ ) about let's say Javascript. You could say that a lot of people that only knows Javascript they don't like computers, because it is so detached from what is going on inside. And the same, once Javascript was famous a lot of people jumped into it, and started creating things, not knowing much about low level stuff, and making terrible things in the eyes of such communities.