Rendered at 20:06:08 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
8by3 1 hours ago [-]
I've got a tangent: I'm done using closed AI.
My last role was HoE after a career as an IC, and honestly, using agents is 95% the same as being an HoE or team lead. Steering the ship, ensuring technical excellence, sometimes fixing dumb bugs, sometimes learning new ways of doing things.
But the whole AI business, especially the closed sort is going to properly screw us all if we let it. At least with open models there is a chance of all of human knowledge being a general good. Not just making the ultra wealthy and powerful, even more so.
zeryx 6 hours ago [-]
So, the existential dread bit - sure. That's a real concern and thing, but let's put that aside for a sec.
When you did math and long division by hand in highschool did you enjoy it? Did you feel like you lost something when you were allowed to use a graphing calculator on a test?
In university I dreaded exams that allowed calculators because I knew how much harder the exam was going to be - that's what genai is today in my experience.
Yes we don't need manual calculators as a profession anymore to land rockets, same as we don't need a programmer to necessarily fiddle around with divs or structs - but the exams are harder now. Now we need to spend our time actually considering trade offs and actually doing software engineering. This is harder and maybe not everyone who's had that job title is particularly good at that - but that's the new skill to get good at, and because the world is under constrained making design and implementation decisions will essentially always need to be done by the person building the thing.
If you liked the programming bit, I'm sorry - I'm personally happy I don't have to remember regex or how to figure out how to center a div or span again.
horticulturist 6 hours ago [-]
Nothing was lost using a graphing calculator because you assume the calculator produces the correct result every time. Imagine if, instead, 90% of the time it was wrong, and you needed to check the result and then take another pull on the slot machine lever.
Centering a div and using regex are basic programming skills that aren’t some kind of difficult exercise or the privilege of a select few. Everyone working (at least on typical business CRUD applications) should know how to use these tools.
If you don’t then I’m sorry - maybe you shouldn’t be a programmer.
zeryx 5 hours ago [-]
I built a ML training framework from scratch in cuda in 2016, for earthquake forecasting - I know quite well how to solve low level kernel optimization problems :)
I'm also regularly a rust developer and I do know REGEX thanks.
Building that project today, would not be impressive or particularly hard. There are great references available today to illustrate how to build that. Sure LLMs aren't 100% accurate, neither is an intern or junior dev you might offload a task to. Part of the job of a software engineer is collaboration. Understanding how to collaborate with teammates (local, AI, outsourced) and how to ship when your teammates make mistakes is quite important to know how to resolve.
That's why CI/CD and TTD exists! If you have a project where correctness is critical, why haven't you written a test to validate that correctness?
marssaxman 1 hours ago [-]
I can't do anything with a div, much less center it - I haven't touched web-tech in twenty years, but I've been a programmer the whole time.
throwawayffffas 5 hours ago [-]
Centering a div was a hack built on hacks based on a spec that was derived from random browser behaviors built over decades. I am super happy I don't have to do that shit again. Both because of flexbox and AI.
But yeah you need to know how to write a regex, if you want to be a programmer.
scotty79 47 minutes ago [-]
> you assume the calculator produces the correct result every time
The calculator itself sure. But the whole process of calculating using calculator? Definitely not 100% always correct. You can miss the buttons, hit the entirely wrong ones, you may put in entirely wrong thing. So if you care about the result you double check, especially if you get something unreasonable. The calculator being perfect is quantitative difference, not qualitative once. The process would be exactly the same if the calculator made 1/100 operations wrong.
jaredcwhite 6 hours ago [-]
Careful, you'll be accused of being an elitist gatekeeper or some sort of nonsense.
But yeah, you're completely right.
impulser_ 13 hours ago [-]
It seems like the problem wasn't the AI, but the fact the guy replaced his brain with an AI. Don't do that and you will be just fine.
firmretention 6 hours ago [-]
Just don't eat too much and you will be just fine.
Just don't use screens so much and you will be just fine.
Once a something becomes a seamlessly integrated and essential part of daily life, "just don't do the bad thing" can become very, very difficult. There was a time where spending all day on the internet was considered maladaptive, and online dating? That's for losers who can't find a match in the real world! Just don't go online so much, and you will be just fine.
throwawayffffas 5 hours ago [-]
I don't think that's fair.
I believe the advice is more like maybe you shouldn't eat the thumbtacks.
Using ai to generate code is one thing asking for book recommendations, career advice, floating business ideas, talking it to it like a friend, running grand ideas past it is not the same.
It's a program that processes and generates words treating it like a person is a one way ticket to depression and delusional thinking.
ColdStream 14 hours ago [-]
A great summary of the issue. It increases productivity but reduces understanding. The flow on from that is a lack of personal engagement and ownership of what is being done. The lack of autonomy can be devastating to a person.
tobinfekkes 14 hours ago [-]
Solid. Well-stated. I'm arriving at a similar point as well.
kaseijin 14 hours ago [-]
I mostly agree but I think most of the issues stemming from it come from management vastly overestimating what AI can actually do vs what it is capable of coupled with people's natural innate laziness and the desire to just "use AI" rather than putting any mental effort into things.
I think AI is useful as a tool in some situations. I use it sometimes as a kind of "google that for me" short cut to save time looking up things I already know. But I don't trust it for things outside my knowledge as I can't be sure if it's correct. I think op biggest issue here is when they start chatting with it normally. AI as a replacement for real human connection is a terrible thing.
One thing I do like using AI for is basically mindless busiwork. Like "I've a list of 1,000 unsorted 12 digit numbers, please count them all, tell me how many repeats, and arrange them in ascending order."
But even then I'm better off programming something if it is more than a one off. Because I still need to check the AI and make sure it's correct.
Only saves a marginal amount of time in certain circumstances.
I really don't like how people just trust what it spits out though without thinking. Eventually someone is gonna do something really stupid with it that causes major issues, like AI powered ATC that ends up crashing planes together. Or something equally grim.
Really I think the tech is interesting and has some really potentially useful applications but right now everyone seems to think it's a magic bullet that solves everything and I don't think really understand the limitations or realities of implementing it to be anything useful.
Personally in my own life I've barely any use for it. Certainly none I'd be willing to pay for. I don't understand how it will be monetized. I basically still treat it like a novelty myself.
I don't much worry about my job getting replaced with AI. It well might but, then I'll just do something else for awhile until they need a specialist who actually understands what is going on without AI (and will charge whatever I want).
f30e3dfed1c9 5 hours ago [-]
"I've a list of 1,000 unsorted 12 digit numbers, please count them all, tell me how many repeats, and arrange them in ascending order."
Gotta say that it is insane to me that people will use tens or hundreds of billions of dollars of hardware to do things like that.
sjsdaiuasgdia 7 hours ago [-]
[dead]
Kim_Bruning 4 hours ago [-]
> but one night when I was sick, I described all my symptoms and it told me to go to the Emergency Department at the hospital.
I had something similar happen, but with a different outcome. One week I noticed I had some mild distortions in one eye I which I thought were a bit funny, one of those things, you know?
No pain or anything, but eh, the next sunday morning I was just messing around, and just for the heck of it, I described the funny distortion thing to Claude Opus...
which proceeded to flip out and told me to call a doctor Right Now.
Clearly Claude was broken, right? And what a weird failure mode! It just kept circling back to telling me to see a doctor "please just call at least!".
I deleted the conversation (something I now regret) and resolved to get on with my day.
But... you know, small doubt having surfaced, I decided to use the expert system provided by my local doctor's practice;
which also told me to call immediately
... and that's how I ended that Sunday morning
in the passenger seat of a Porsche
racing down the autobahn to a nearby hospital.
Long story short: I think Claude gets some credit for preventing me from going partially blind. (and don't mess around with your eyesight.)
throwaway985732 6 hours ago [-]
I can't say I feel the same.
> Well, it made me lazy. It made me stop caring. It made me a worse programmer. It made me depressed. Because I stopped doing the hard work, I stopped learning, I stopped growing, I stopped being the one making the software.
After burning out hard a few years ago and doing this job with literally nothing to show for close to 20 years.I have to say I was already all of those things years ago before AI came along. So the idea of having something do the coding for me is actually refreshing.
Code is stupid the whole idea is to write less of it.
xyzsparetimexyz 7 hours ago [-]
Alright well see youu again next month I guess
supermarvin 14 hours ago [-]
[dead]
scotty79 30 minutes ago [-]
tl;dr
Psych issue, he got lazy, detached and depressed. He blames AI for it so decided to set it aside.
If I were to guess his depression is due to something physical (my belief is that it always is) and all AI did was to allow him to be productive a employee despite deepening depression slash burnout.
My last role was HoE after a career as an IC, and honestly, using agents is 95% the same as being an HoE or team lead. Steering the ship, ensuring technical excellence, sometimes fixing dumb bugs, sometimes learning new ways of doing things.
But the whole AI business, especially the closed sort is going to properly screw us all if we let it. At least with open models there is a chance of all of human knowledge being a general good. Not just making the ultra wealthy and powerful, even more so.
When you did math and long division by hand in highschool did you enjoy it? Did you feel like you lost something when you were allowed to use a graphing calculator on a test?
In university I dreaded exams that allowed calculators because I knew how much harder the exam was going to be - that's what genai is today in my experience.
Yes we don't need manual calculators as a profession anymore to land rockets, same as we don't need a programmer to necessarily fiddle around with divs or structs - but the exams are harder now. Now we need to spend our time actually considering trade offs and actually doing software engineering. This is harder and maybe not everyone who's had that job title is particularly good at that - but that's the new skill to get good at, and because the world is under constrained making design and implementation decisions will essentially always need to be done by the person building the thing.
If you liked the programming bit, I'm sorry - I'm personally happy I don't have to remember regex or how to figure out how to center a div or span again.
Centering a div and using regex are basic programming skills that aren’t some kind of difficult exercise or the privilege of a select few. Everyone working (at least on typical business CRUD applications) should know how to use these tools.
If you don’t then I’m sorry - maybe you shouldn’t be a programmer.
Building that project today, would not be impressive or particularly hard. There are great references available today to illustrate how to build that. Sure LLMs aren't 100% accurate, neither is an intern or junior dev you might offload a task to. Part of the job of a software engineer is collaboration. Understanding how to collaborate with teammates (local, AI, outsourced) and how to ship when your teammates make mistakes is quite important to know how to resolve.
That's why CI/CD and TTD exists! If you have a project where correctness is critical, why haven't you written a test to validate that correctness?
But yeah you need to know how to write a regex, if you want to be a programmer.
The calculator itself sure. But the whole process of calculating using calculator? Definitely not 100% always correct. You can miss the buttons, hit the entirely wrong ones, you may put in entirely wrong thing. So if you care about the result you double check, especially if you get something unreasonable. The calculator being perfect is quantitative difference, not qualitative once. The process would be exactly the same if the calculator made 1/100 operations wrong.
But yeah, you're completely right.
Just don't use screens so much and you will be just fine.
Once a something becomes a seamlessly integrated and essential part of daily life, "just don't do the bad thing" can become very, very difficult. There was a time where spending all day on the internet was considered maladaptive, and online dating? That's for losers who can't find a match in the real world! Just don't go online so much, and you will be just fine.
I believe the advice is more like maybe you shouldn't eat the thumbtacks.
Using ai to generate code is one thing asking for book recommendations, career advice, floating business ideas, talking it to it like a friend, running grand ideas past it is not the same.
It's a program that processes and generates words treating it like a person is a one way ticket to depression and delusional thinking.
I think AI is useful as a tool in some situations. I use it sometimes as a kind of "google that for me" short cut to save time looking up things I already know. But I don't trust it for things outside my knowledge as I can't be sure if it's correct. I think op biggest issue here is when they start chatting with it normally. AI as a replacement for real human connection is a terrible thing.
One thing I do like using AI for is basically mindless busiwork. Like "I've a list of 1,000 unsorted 12 digit numbers, please count them all, tell me how many repeats, and arrange them in ascending order."
But even then I'm better off programming something if it is more than a one off. Because I still need to check the AI and make sure it's correct.
Only saves a marginal amount of time in certain circumstances.
I really don't like how people just trust what it spits out though without thinking. Eventually someone is gonna do something really stupid with it that causes major issues, like AI powered ATC that ends up crashing planes together. Or something equally grim.
Really I think the tech is interesting and has some really potentially useful applications but right now everyone seems to think it's a magic bullet that solves everything and I don't think really understand the limitations or realities of implementing it to be anything useful.
Personally in my own life I've barely any use for it. Certainly none I'd be willing to pay for. I don't understand how it will be monetized. I basically still treat it like a novelty myself.
I don't much worry about my job getting replaced with AI. It well might but, then I'll just do something else for awhile until they need a specialist who actually understands what is going on without AI (and will charge whatever I want).
Gotta say that it is insane to me that people will use tens or hundreds of billions of dollars of hardware to do things like that.
I had something similar happen, but with a different outcome. One week I noticed I had some mild distortions in one eye I which I thought were a bit funny, one of those things, you know?
No pain or anything, but eh, the next sunday morning I was just messing around, and just for the heck of it, I described the funny distortion thing to Claude Opus... which proceeded to flip out and told me to call a doctor Right Now.
Clearly Claude was broken, right? And what a weird failure mode! It just kept circling back to telling me to see a doctor "please just call at least!".
I deleted the conversation (something I now regret) and resolved to get on with my day.
But... you know, small doubt having surfaced, I decided to use the expert system provided by my local doctor's practice;
which also told me to call immediately
... and that's how I ended that Sunday morning
Long story short: I think Claude gets some credit for preventing me from going partially blind. (and don't mess around with your eyesight.)> Well, it made me lazy. It made me stop caring. It made me a worse programmer. It made me depressed. Because I stopped doing the hard work, I stopped learning, I stopped growing, I stopped being the one making the software.
After burning out hard a few years ago and doing this job with literally nothing to show for close to 20 years.I have to say I was already all of those things years ago before AI came along. So the idea of having something do the coding for me is actually refreshing.
Code is stupid the whole idea is to write less of it.
Psych issue, he got lazy, detached and depressed. He blames AI for it so decided to set it aside.
If I were to guess his depression is due to something physical (my belief is that it always is) and all AI did was to allow him to be productive a employee despite deepening depression slash burnout.