AI - The Good The Bad The Ugly

[RustVendor]

I selected “Good” but it’s closer to “Both” for me. I just didn’t like the “hate it” framing.

I use it enthusiastically and it has helped me solve larger, more complex problems at a much faster rate.

Like most things in life it’s just a tool and how we use it determines whether it’s ultimately helpful or destructive.

By far, the biggest problem to solve in tech is stability. It's wild that we have come this far and computers still barely work, every time it comes close new "features" are added which increase complexity and we always exist on the bleeding edge. That's why I critique vibe coding, we have plenty of ideas for new features but they aren't worth a damn if they just make things work poorly. The underlying tech is what allows features to be stable. It's the difference between Amazon AWS and Microsoft products. Amazon has always prioritized stability whereas MS is constantly adding broken features.

 

[john_marsten_1873]

By far, the biggest problem to solve in tech is stability. It's wild that we have come this far and computers still barely work, every time it comes close new "features" are added which increase complexity and we always exist on the bleeding edge. That's why I critique vibe coding, we have plenty of ideas for new features but they aren't worth a damn if they just make things work poorly. The underlying tech is what allows features to be stable. It's the difference between Amazon AWS and Microsoft products. Amazon has always prioritized stability whereas MS is constantly adding broken features.

The stability vs. innovation problem will always exist. I’ve iterated between what I like to frame as technologist vs. builder often, where I’m hyper focused on the tools vs. product/outcome: clean code, low tech-debt vs. spending more time on features that customers/users are asking for.

And of course, I’ve been on the side of neglecting tech-debt cleanup has directly impacted our ability and/or speed to release important features that depended on a new customer contract or renewal. Also, the dreaded “well the prototype works so I don’t think we should just throw away that work.” from the execs…

Right now, the market is favoring builders: finding meaningful use cases for AI. And the reality is that it can be fruitful for both engineers and vibe coders. There’s a lot of great ideas out there that previously, folks had to spend $5-10k to hire an engineer just to prototype. Now, there’s never been a better opportunity for non-technical folks to build a proof of concept and partner with engineers to build something that could scale to at least the first 100k users.

AWS won because of their innovation and aggressiveness toward capturing market share. Too many people were comfortable with “we’re a Microsoft shop” and Microsoft wasn’t going to rush folks toward cloud infra and compute.

The reality that is there has to be enough stability to drive enough innovation so you don’t get disrupted. These days, I regularly push back against tech debt cleanup or new tech proposals that don’t clearly unblock/unlock product features, or meaningfully and measurably improve development speed and productivity.

 

[ikeo1]

The stability vs. innovation problem will always exist. I’ve iterated between what I like to frame as technologist vs. builder often, where I’m hyper focused on the tools vs. product/outcome: clean code, low tech-debt vs. spending more time on features that customers/users are asking for.

And of course, I’ve been on the aide of neglecting tech cleanup has directly impacted our ability and/or speed to release important features that depended on a new customer contract and renewal. Also, the dreaded “well the prototype works so I don’t think we should just throw away that work.” from the execs.

Right now, the market is favoring builders: finding meaningful use cases for AI. And the reality is that it can be fruitful for both engineers and vibe coders. There’s a lot of great ideas out there that previously, folks had to spend $5-10k to hire an engineer just to prototype. Now, there’s never been a better opportunity for non-technical folks to build a proof of concept and parter up to build something that could scale to at least the first 100k users.

AWS won because of their innovation and aggressiveness toward capturing market-share. Too many people were comfortable with “we’re a Microsoft shop” and Microsoft wasn’t going to rush folks toward cloud infra and compute.

The reality that there has to be enough stability to drive enough innovation so you don’t get disrupted. These days, I regularly push back against tech debt cleanup or new tech, proposals that don’t clearly unblock/unlock product features or meaningfully and measurably improve development speed and productivity.

There’s always a balance. It’s not just in tech, but any product in general. If it’s useful, it’ll grow. It’s just easier to get it built now. If it’s not, it’s garbage, which most people are producing because now everybody has access to the tool, and you get a lot of trash. That doesn’t mean the tech is bad or the tech fault. You can’t blame the technology and people’s understanding of it when the average education is going down.

 

[RustVendor]

The stability vs. innovation problem will always exist. I’ve iterated between what I like to frame as technologist vs. builder often, where I’m hyper focused on the tools vs. product/outcome: clean code, low tech-debt vs. spending more time on features that customers/users are asking for.

And of course, I’ve been on the aide of neglecting tech cleanup has directly impacted our ability and/or speed to release important features that depended on a new customer contract and renewal. Also, the dreaded “well the prototype works so I don’t think we should just throw away that work.” from the execs.

Right now, the market is favoring builders: finding meaningful use cases for AI. And the reality is that it can be fruitful for both engineers and vibe coders. There’s a lot of great ideas out there that previously, folks had to spend $5-10k to hire an engineer just to prototype. Now, there’s never been a better opportunity for non-technical folks to build a proof of concept and parter up to build something that could scale to at least the first 100k users.

AWS won because of their innovation and aggressiveness toward capturing market-share. Too many people were comfortable with “we’re a Microsoft shop” and Microsoft wasn’t going to rush folks toward cloud infra and compute.

The reality that there has to be enough stability to drive enough innovation so you don’t get disrupted. These days, I regularly push back against tech debt cleanup or new tech, proposals that don’t clearly unblock/unlock product features or meaningfully and measurably improve development speed and productivity.

That's the thing though, stability is a feature in and of itself. As someone forced to use Microsoft products, the fact that even their basic office tools like Word and Even F-ing Outlook are constantly bugging out is a lot more meaningful than the fact they they integrated copilot. most work isn't using new features it's using the core functions. Like why are we taking about AI when Word can barely support an online spreadsheet. Why is the answer so much of the time to just copy and paste into a new document? That's literally what MS support tells you because they don't know either. Everything from our phones to what should be a fairly basic email reader is still constantly bugging. I literally don't need any new features just make the core functions work.

 

[mblat1968]

There’s always a balance. It’s not just in tech, but any product in general. If it’s useful, it’ll grow. It’s just easier to get it built now. If it’s not, it’s garbage, which most people are producing because now everybody has access to the tool, and you get a lot of trash. That doesn’t mean the tech is bad or the tech fault. You can’t blame the technology and people’s understanding of it when the average education is going down.

I don't disagree. However... I don't think you can blame luck of understanding of technology on bad education.
I think technology has moved beyond understanding of an average person regardless of what education they will get.

 

[RustVendor]

I don't disagree. However... I don't think you can blame luck of understanding of technology on bad education.
I think technology has moved beyond understanding of an average person regardless of what education they will get.

I think he's talking about actual programmers who are less and less capable these days, not laymen and product users per se

 

[mblat1968]

I think he's talking about actual programmers who are less and less capable these days, not laymen and product users per se

Well. Quality of software engineers have been declining for decades.

My horror story I think is pre COVID. We interviewed one recent MS is CS who was working as a teacher assistant .

She could not write few lines of code if her life dependent on that. In language of her choice.

 

[Darqhelmet]

I think those in the Bay Area and other metro areas could be and are being effected first. Microsoft is offering buy out package for 60,000 of their 240,000 employees (25%) of their workforce. I'm not sure those in middle America will be as affected in the near term until AI Robots take their manual labor type jobs. Ultimately all of the US (and the World) will be affected. The challenge for the Governments of the World will be what/how to deal with mass unemployment in the next 10 years. Elon says "Robots will do everything freeing up mankind to pursue their interests and only work if they want to". I don't see that happening!

yeah, not what I am talking about. the data centers, noise, power cost, property value loss, poisoning of neighborhoods from jet engines running 24/7. not happening in the Bay or cities in general. Tech people getting laid off is nothing new. My uncle spent his entire life having, not having jobs on a every fre year cycle. He was a network architect for IBM, McDonald, DoD, etc

 

[ikeo1]

Well. Quality of software engineers have been declining for decades.

My horror story I think is pre COVID. We interviewed one recent MS is CS who was working as a teacher assistant .

She could not write few lines of code if her life dependent on that. In language of her choice.

Unfortunately, I’m in this group but conceptually speaking I understand entirely what I want to happen. The only issue is getting caught up in syntax and understanding object orientation while keeping mind of the entire picture. I’m always impressed by the people who are able to do it. I have limited capabilities, but I try to use them as best I can. now that I have extra tools, I’m using them as far as I can take them. It’s actually quite amazing. I’m frequently discovering new capabilities I’m trying to put them into practice. The hard part is structuring everything so that it’s consistent and reliable. When you make changes one of the worst things that people do is just give it a task and let it run forever. I think that’s the biggest issue right now that causes people pain and the most resource strain the people who don’t give it good guidance the systems will turn and loop over and over causing the drain of resources. If you’re efficient, then it gets to the end goal quite quickly, but people who aren’t able to articulate their goals or processes and even structure eat inadvertently cause extra environmental damage.

I do think that eventually they’ll pull the rug on us. A lot of this the utilization will go down for the price, which is why I’ve invested in local LLMs. They can’t do all the work, but they do reduce my need on the frontier models. I can also subsidize my power usage with solar since I run on Macs they use significantly less power.

Overall, he definitely reduces the number of jobs, but the unreliability is predictable. With people you never know when they’re gonna run away. The people who are good at their craft will do even better. The people who are so so are likely to get pushed out and have to find something else to do.

 

[john_marsten_1873]

That's the thing though, stability is a feature in and of itself. As someone forced to use Microsoft products, the fact that even their basic office tools like Word and Even F-ing Outlook are constantly bugging out is a lot more meaningful than the fact they they integrated copilot. most work isn't using new features it's using the core functions. Like why are we taking about AI when Word can barely support an online spreadsheet. Why is the answer so much of the time to just copy and paste into a new document? That's literally what MS support tells you because they don't know either. Everything from our phones to what should be a fairly basic email reader is still constantly bugging. I literally don't need any new features just make the core functions work.

Shows both how lucky and out-of-touch I am. I don’t regularly use ANY MS products these days except for checking some old emails, once in a blue moon.

But I do think Microsoft is a rather unique case as their systems have been around so long and still power so many, today.

Especially for things like Office and Sharepoint, they should certainly optimize more on the side of stability. At the same time, they’ve worked on getting companies so heavily dependent on them that it would take A LOT for folks to switch GSuite or standalone products. If they’re out of SLO targets they’ll just discount contract renewals or otherwise to keep people locked-in.

 

[RustVendor]

Shows both how lucky and out-of-touch I am. I don’t regularly use ANY MS products these days except for checking some old emails, once in a blue moon.

But I do think Microsoft is a rather unique case as their systems have been around so long and still power so many, today.

Especially for things like Office and Sharepoint, they should certainly optimize more on the side of stability. At the same time, they’ve worked on getting companies so heavily dependent on them that it would take A LOT for folks to switch GSuite or standalone products. If they’re out of SLO targets they’ll just discount contract renewals or otherwise to keep people locked-in.

Sharepoint is such a piece of garbage it's the bane of my existence. MS survived the antitrust lawsuits of the 90s and we see the product of monopolization, it ain't good for anyone, I don't even think it's good for MS ultimately.

 

[mblat1968]

Unfortunately, I’m in this group but conceptually speaking I understand entirely what I want to happen. The only issue is getting caught up in syntax and understanding object orientation while keeping mind of the entire picture. I’m always impressed by the people who are able to do it. I have limited capabilities, but I try to use them as best I can. now that I have extra tools, I’m using them as far as I can take them. It’s actually quite amazing. I’m frequently discovering new capabilities I’m trying to put them into practice. The hard part is structuring everything so that it’s consistent and reliable. When you make changes one of the worst things that people do is just give it a task and let it run forever. I think that’s the biggest issue right now that causes people pain and the most resource strain the people who don’t give it good guidance the systems will turn and loop over and over causing the drain of resources. If you’re efficient, then it gets to the end goal quite quickly, but people who aren’t able to articulate their goals or processes and even structure eat inadvertently cause extra environmental damage.

I do think that eventually they’ll pull the rug on us. A lot of this the utilization will go down for the price, which is why I’ve invested in local LLMs. They can’t do all the work, but they do reduce my need on the frontier models. I can also subsidize my power usage with solar since I run on Macs they use significantly less power.

Overall, he definitely reduces the number of jobs, but the unreliability is predictable. With people you never know when they’re gonna run away. The people who are good at their craft will do even better. The people who are so so are likely to get pushed out and have to find something else to do.

Well... You are not MS in CS I assume....

Yes that is where AI gives the most gaines. Make it work and I don't really care how. However there are a lot of places where how matters. In those places even best AI models will tell you that it should not be used without extreme prejudice from human reviewers.

 

[mblat1968]

Sharepoint is such a piece of garbage it's the bane of my existence. MS survived the antitrust lawsuits of the 90s and we see the product of monopolization, it ain't good for anyone, I don't even think it's good for MS ultimately.

I don't know if it is result of monopoly.....
But sure

Sharepoint and MS Teams in general is big pile of

 

[Darqhelmet]

I don't know if it is result of monopoly.....
But sure

Sharepoint and MS Teams in general is big pile of

Teams can burn in hell.

 

[RustVendor]

I don't know if it is result of monopoly.....
But sure

Sharepoint and MS Teams in general is big pile of

If they didn't have a monopoly we would use something else or at the very least there would be some amount of competition and should force MS to make a better product. Even just their operating system sucks, if it's not the result of a monopoly then why does it suck so hard?