Tech Thread

[zeke]

It’s not useful if you already know everything. But for everyone else, it’s quite helpful.

For example though - whenever i check the actual links in google's AI Overview i realize that about half the time the AI is completely wrong about what the link says. Just plain wrong. That's from simple stuff like asking for an NBA player's combine measurements or for something a bit more complicated like an Ontario legal statute question. It's legit terrible.

Private models may or may not be better but i just had a horrendous experience with ChatGPT where i thought something it gave me sounded good and just needed some editing and i wasted hours and hours editing before i realized it was actually fundamentally wrong and had to do it all from scratch again.

 

[CH1]

Private models may or may not be better but i just had a horrendous experience with ChatGPT where i thought something it gave me sounded good and just needed some editing and i wasted hours and hours editing before i realized it was actually fundamentally wrong and had to do it all from scratch again.

That legit sucks. I mostly use GPT but if I’m unsure about it, I will double check with Claude or Perplexity.

 

[Preston]

I think it's great. Flawed and often shit if you're dumb, but if you know how to use it (much like knowing how to do a Google search efficiently) you can get a wealth of knowledge at record speeds. Incredible tool, completely Revolutionary, but the question will always be how profitable it can realistically become.

My bet is, as is usually the case with groundbreaking tech, the path of least resistance, least red tape, and probably most profitability, will come through AI porn.

 

[zeke]

I love tech and embrace it but this one simply hasn't worked for me at all and I suspect most people are getting really bad results without realizing it.

FYI: that thing i was just talking about that ChatGPT really screwed me on? literally 5 minutes ago i just got an absolute homerun response from what i did after throwing out the AI result completely and starting again from scratch. absolute homerun. thank god i threw out the AI slop.

 

[Preston]

I spend most of my days on Excel and coding, and for those two items, it has absolutely changed the game. I do things that I simply never would have been able to do before. Or if I did, it would have taken several hours. It's a cheat code for what I use it for.

 

[MindzEye]

I love tech and embrace it but this one simply hasn't worked for me at all and I suspect most people are getting really bad results without realizing it.

FYI: that thing i was just talking about that ChatGPT really screwed me on? literally 5 minutes ago i just got an absolute homerun response from what i did after throwing out the AI result completely and starting again from scratch. absolute homerun. thank god i threw out the AI slop.

How are you at prompting? I find a huge difference in quality of response compared to my early attempts to use it. I find that I get a lot out of it when I'm willing to sit down and have a properly structured prompt argument with it, challenge it on things that seem off in it's answers, etc. Context, use case, desired type of output, and as specific as possible. Then dig through it's answer looking for inconsistencies and bullshit, challenge it on those things but I keep a notepad going to C&P anything good as we go, because each new iteration...is a new iteration, regardless of how well it tries to pass it off as a continued chain of thought.

 

[MindzEye]

I spend most of my days on Excel and coding, and for those two items, it has absolutely changed the game. I do things that I simply never would have been able to do before. Or if I did, it would have taken several hours. It's a cheat code for what I use it for.

Yeah, same. Zeke was right earlier when he mentioned that it's useful for fairly technical things that are either right/wrong. Codes/scripts, step by step guides on obscure software commands, etc. Because the shit either works the way it says it does and you're golden, or it doesn't and you challenge it until you ask the question the right way, or it stops fucking hallucinating (hard to tell which is which).

I've spent 2-3 hours trying to get it to debug code that it's built for me, and continued to fail to fix it, forcing me to torch the entire thing and start over, but it's also helped me do shit in 10 minutes that I simply wouldn't have been capable of on my own and learning it would have taken days or weeks.

 

[mbow30]

Has it dawned on anybody that maybe Zeke is an Ai creation?

 

[zeke]

How are you at prompting? I find a huge difference in quality of response compared to my early attempts to use it. I find that I get a lot out of it when I'm willing to sit down and have a properly structured prompt argument with it, challenge it on things that seem off in it's answers, etc. Context, use case, desired type of output, and as specific as possible. Then dig through it's answer looking for inconsistencies and bullshit, challenge it on those things but I keep a notepad going to C&P anything good as we go, because each new iteration...is a new iteration, regardless of how well it tries to pass it off as a continued chain of thought.

Yeah I definitely haven't spent the time developing my prompting skills. But in this case i actually inherited the bulk of output from a smart educated professional who uses ChatGPT every day and he had spent time prompting and discussing it with the AI, and this was something of great personal importance to him. I probably could have improved the output with my personal expertise but at that point i'm wondering if we're spending more time prompting than i would need to do it from scratch in the first place.

 

[zeke]

I spend most of my days on Excel and coding, and for those two items, it has absolutely changed the game. I do things that I simply never would have been able to do before. Or if I did, it would have taken several hours. It's a cheat code for what I use it for.

yeah this sounds right.

 

[Preston]

Yeah, same. Zeke was right earlier when he mentioned that it's useful for fairly technical things that are either right/wrong. Codes/scripts, step by step guides on obscure software commands, etc. Because the shit either works the way it says it does and you're golden, or it doesn't and you challenge it until you ask the question the right way, or it stops fucking hallucinating (hard to tell which is which).

I've spent 2-3 hours trying to get it to debug code that it's built for me, and continued to fail to fix it, forcing me to torch the entire thing and start over, but it's also helped me do shit in 10 minutes that I simply wouldn't have been capable of on my own and learning it would have taken days or weeks.

It's no different than a Google search, it requires a prerequisite amount of intelligence to do that efficiently too. Many people use AI as a replacement for their brain, which ain't it. Every fucking twitter reply is @grok what does this mean? Or @grok is this true? No, it doesn't replace using your own critical thinking. It's a tool to help enhance your knowledge and capabilities, and imo a really good one for certain uses.

But I still say the end game is porn, not curing cancer.

 

[zeke]

Yeah, same. Zeke was right earlier when he mentioned that it's useful for fairly technical things that are either right/wrong. Codes/scripts, step by step guides on obscure software commands, etc. Because the shit either works the way it says it does and you're golden, or it doesn't and you challenge it until you ask the question the right way, or it stops fucking hallucinating (hard to tell which is which).

I've spent 2-3 hours trying to get it to debug code that it's built for me, and continued to fail to fix it, forcing me to torch the entire thing and start over, but it's also helped me do shit in 10 minutes that I simply wouldn't have been capable of on my own and learning it would have taken days or weeks.

this is probably the best way to use it - for something you might literally just not be able to do otherwise (i.e. write code) and where the output is simply a right/wrong type output.

 

[zeke]

It's no different than a Google search, it requires a prerequisite amount of intelligence to do that efficiently too. Many people use AI as a replacement for their brain, which ain't it. Every fucking twitter reply is @grok what does this mean? Or @grok is this true? No, it doesn't replace using your own critical thinking. It's a tool to help enhance your knowledge and capabilities, and imo a really good one for certain uses.

But I still say the end game is porn, not curing cancer.

but the google AI overview keeps reading my search results wrong for me. like blatantly wrong. it's a problem.

Curing cancer, though, is still the kind of thing that i think AI is worth dreaming on for.

 

[axlsalinger]

yeah i use it quite a lot, and it's why I'm still relatively bullish on AI in general. It can be very useful in many different ways. I will say that every single day I use it for a decent amount of time, at some point I'm always (a) astounded at how good a response was, and (b) how horrifically stupid a response was.

We are still in early days... there's no question in my mind that it's here to stay and will continue to improve over time.

 

[zeke]

AI porn loses its luster pretty quick tbh.

 

[Preston]

but the google AI overview keeps reading my search results wrong for me. like blatantly wrong. it's a problem.

Curing cancer, though, is still the kind of thing that i think AI is worth dreaming on for.

I guess we're coming at this from different angles based on how we use it. I'm mostly just agreeing with mindzeyes claim about the value in solid prompts and understanding its limitations at the same time. For coding, you learn quickly how important this is because even for that it's not like it'll get it right the first time. A lot of it is troubleshooting with the agent after its first attempt and eventually coming towards a solution. Sometimes even presenting your own ideas on how a broken code can be fixed. And in the process, you even learn things yourself along the way, I've become a waaaaay better coder since this shit was introduced. People need to think of it as a tool and not a human replacement. At least at this stage.

Curing cancer is a valid goal, I just don't think that'll be the goal of these corporations since there are paths of far less resistance. And we're way further away from accomplishing that goal than Elon "everyone's job will be replaced by AI in 3 years" Musk thinks.

 

[axlsalinger]

AI porn loses its luster pretty quick tbh.

VR.

 

[MindzEye]

this is probably the best way to use it - for something you might literally just not be able to do otherwise (i.e. write code) and where the output is simply a right/wrong type output.

Easily the best use case. It's really only useful for online shitpost research otherwise.

 

[mbow30]

AI porn loses its luster pretty quick tbh.

your prompts must be shit.

 

[Preston]

Easily the best use case. It's really only useful for online shitpost research otherwise.

ChatGPT said:
Here’s what I found out (and what remains uncertain) about “Preston” on Forum Ice (forumice.com) — the hockey forum user.

What seems to be true about “Preston”
On Forum Ice, there is a frequent user going by the name “Preston” — often also shown as “Preston (MBow30 alt account)”.
Forum Ice
+2
Forum Ice
+2

“Preston” posts about hockey topics (e.g. discussing the Toronto Maple Leafs, the NHL, prospects, goaltending, etc.).
Forum Ice
+2
Forum Ice
+2

They seem to engage often — participating in “Around the League” threads, game-day threads, off-season threads, and general hockey-related discussion.
Forum Ice
+2
Forum Ice
+2

From a thread in 2025-2026 off-season there’s a post by Preston complaining about the state of the team and with strong opinions.
Forum Ice

What’s murky / contested
There is ongoing suspicion among other users that “Preston” might not be one real person — but perhaps using alternate accounts (alts). For example, someone says “looks like presto got bored again” and “Fucking Preston” in a context that suggests trolling or alt-account behaviour.
Forum Ice
+2
Forum Ice
+2

The alias “MBow30 alt account” attached to Preston raises questions — it suggests that “Preston” may at times post as a secondary identity, or that other users suspect “Preston” of using multiple identities.
Forum Ice
+2
Forum Ice
+2

There is no publicly reliable profile identifying “Preston” off the forum (no real name, bio, location, etc.). So it’s difficult — perhaps impossible — to know exactly who “Preston” is in real life.

Community perception & controversies
The community seems somewhat divided or at least wary about “Preston.” Some users appear to treat him as a valid (if strong-voiced) hockey commenter. Others treat him skeptically — suspecting alt-account usage, trolling, or looking down on “Preston” when his posts are controversial or provocative.
Forum Ice
+2
Forum Ice
+2

One user comment from a thread reads:

“Looks like presto got bored again” — implying that “Preston” might jump around between accounts when bored.
Forum Ice

Another thread shows “Preston” making strong opinions about team performance and hockey analytics:

“…I don’t like any qualitative analysis really, even from ex-players… Anything.”
Forum Ice

All told — “Preston” seems to be a longtime, opinionated, somewhat polarizing poster on Forum Ice.

What you should keep in mind if you reference information from “Preston”
Because of the suspicion around alts and the lack of verifiable identity, anything “Preston” says should be taken as a forum user’s opinion, not an objective fact.

The tone and style may change — could be serious hockey takes, sarcasm, trolling, or shifting between accounts — so context matters.

“Preston” appears to be active mostly in Leafs- and NHL-related threads, and often represents a certain mindset (skepticism toward “media narratives”, preference for analytics, critical of “hockey culture” tropes).