Yeah there’s a difference between dumping all the instructions into ChatGPT and hoping it one shots the entire thing and asking it a niche question that it might know the answer to so I don’t have to go digging in man pages for 20 minutes to come up with the command I need. I’m perfectly ok with and would encourage the latter most of the time since that’s not much different than asking a TA for help on something after already trying. “Using it for courses” can mean lots of different shit
I’d argue that, used properly, using genAI can be MORE intellectually stimulating than other means of learning. You can basically probe the world’s knowledge with pretty good reliability on-demand, so the feedback loop for learning gets much much shorter. Again I’m not arguing that having it write all your essays or code is good for learning, but asking targeted questions definitely can be.
I think having a machine spit all of it out to you rather than taking the time to seek it out yourself and read what it tells you from the original sources is far more intellectually stimulating. Gen AI just fast scans the internet and regurgitates a facsimile of what an original source initially had to say, often with inaccuracies baked into it. In my opinion that’s far less valuable than people make it out to be. I see the value in AI tech in some fields, such as medical study and…
Have you ever needed to write a bash script that would render Kubernetes manifests for the same services located in 30 different files but using two different versions of your microservices’ helm chart, generate the diffs between them, then compare those to a reference to see if anything will break in an upgrade? Yeah, me neither until last week. And even though I’m sure I could’ve done it given 30 minutes or so, Copilot got me to something that I got working in 2.
I know how to do it, I don’t feel like taking the time to write the code and inevitably debug my mistakes when a machine can get me 95% of the way there and I just have to do a bit of debugging. And don’t you think to some extent, if a machine can do my work in a fraction of the time, learning how to use that machine to increase my productivity ought to be a good skill to have?
Except it’s not YOUR productivity. The machine is doing the work and production. At a certain point, what value do you bring to the position? Why even have you around in the company if all you need is to pump the prompt into the machine? Why pay you as much for your position if that’s all you’re doing within it? Do you see what I’m getting at here
Using AI as a tool to further your own work and ingenuity is one thing, using it in place of it is another. It’s part of why some people in industries like yours and mine are already losing their jobs to this kind of thing. When you fail to see the value in having human minds and human creation behind work, you end up kicking those humans to the curb to save money and cut out the middleman.
lEveryone learns in different ways. I am cognizant of AI hallucinations and I’m not going to just take everything it says at face value. But I think to some extent you’re missing what I’m saying: not every question has an answer that exists somewhere on the public Internet. Sometimes it’s really useful to ask niche questions that I couldn’t just Google and have AI give me an idea. It doesn’t necessarily need to be a replacement, but a complement to traditional methods
Did tractors not make farmers more productive? Did trucks not make mailmen more productive? See anyone can have ChatGPT generate code for them, not everyone can have ChatGPT generate the code that’s needed. I KNEW what the script needed to do. I GAVE Copilot the necessary information. I still FIXED its mistakes. So yeah, it is MY productivity. If companies can just pay engineers less and fire them all because of this whole AI thing, why haven’t they already? It’s because they can’t
The idea that anyone qualified to work in tech is losing their job to AI is a complete sham. Conveniently almost all the companies that tell you they’re doing that also sell AI products. In reality they all overextended themselves during Covid and now had to either offshore jobs to India or cut them entirely
Except a tractor and a truck doesn’t complete the work for you, it merely expedites a portion of the work. You’re comparing a machine that produces answers to equations to ones that simply allow physical labor to progress at a higher rate. Those machines do things a human being cannot. You yourself admitted you can generate the code ChatGPT does. A human being cannot do what a truck or tractor can, but a human can do what ChatGPT does. Again, when it comes to employers looking to cut corners…
…at a certain point they’re going to start reconsidering how much labor is being put into a position if a machine is expediting a major process and how much that position is worth, how much salary should be put into it. The same thing is happening in many industries. Breaking news and brief writers are being laid off at major publications because ChatGPT can do it passably well.
And ChatGPT didn’t complete the work for me. It automated one part of my task that also involved reading documentation and specifications, communicating with stakeholders, monitoring for incidents, etc. It generated me a mostly working script (that again, I still needed to fix) that helped me expedite the step of verifying that an upgrade wouldn’t break anything when deployed. So it expedited a portion of my work. My job isn’t just writing code all day
And wtf do you mean a tractor and a truck do things people can’t? How do you think anyone farmed or delivered mail before those things existed? Sure the methods are obviously very different, but the end result of planted fields/harvested crops or delivered mail happened prior to the machinery. Those machines made the humans more productive at their jobs, just like AI that can sometimes give me good code does
I mean to be completely frank I think there’s a pretty big difference. If a news article is written “passably well,” then that’s kinda just it. The publication still makes money. Code can’t be written “passably well” or all your customer data will be in Russia before you can count to ten