2026 is the Year of AI

  • Thread starter Thread starter PityOnU
  • Start date Start date
  • Views Views 2,464
  • Replies Replies 35
  • Likes Likes 5

PityOnU

Well-Known Member
Member
Joined
Jul 5, 2012
Messages
1,223
Reaction score
784
Trophies
1
XP
1,853
Country
United States
Just figured I'd post about this here, because I'm now seeing huge shifts on my team and in the rest of the industry in general, and the reality of what's happening in tech is no longer aligned with public opinion. Ofc, I'm talking about AI.

Just as a preface, a little about me is that I have a doctorate in Computer Engineering, and I have about a decade of experience working across companies including Intel, Microsoft (Xbox), and Meta (Oculus). I've also been a huge AI skeptic. Out of 40+ people in my research lab in grad school, I was the only one NOT working on AI/ML. With that out of the way...

2026 is the year of AI. It's not hype anymore. It's here. Is it perfect? No. But the absolute latest models (as in, within the last few months) are now "good enough." It is genuinely useful, mostly correct, and orders of magnitude faster than humans at tasks like research and software development.

As of the past few weeks, no one on my team writes code themselves anymore. This is also the pattern seen by any colleagues I have across the industry. If you work in tech, as a software engineer, the medium-term future of your job is piloting 3-5 AI agents in parallel like a coked-out, somehow-nerdier Shinji.

Just figured I would fire the alert flares here as the general public (and especially gamers) are still violently anti-AI. But the reality is that the genie is now out of the bottle and the tools are so good that you now are dumb if you don't use it. It makes things that were previously impossible simply due to lack of RoI completely doable. I will point to the recent Animal Crossing PC port as an example of this.

I guess what I'm trying to say is, don't immediately shit all over devs who use it (homebrew or indie, especially), because you're going to be scaring away a lot of cool projects for no reason.
 
I guess what I'm trying to say is, don't immediately shit all over devs who use it (homebrew or indie, especially), because you're going to be scaring away a lot of cool projects for no reason.
Watch me do it anyway.
 
I guess what I'm trying to say is, don't immediately shit all over devs who use it (homebrew or indie, especially), because you're going to be scaring away a lot of cool projects for no reason.
No, I am going to shit on them more. Fuck them
 
Well, this thread is a good example of what I mean. But let us consider an example that I myself went through the other day. Once again, related to the Animal Crossing PC port:

I own an Anbernic RG40XX H device. I wanted to get the Animal Crossing PC port running on it. For those unfamiliar with development, here's what that would have looked like:

  • Finding out what the AC decompilation project actually entailed. What code was there. What wasn't. What libraries did it need?
  • Understanding the Gamecube SDK, which is probably used heavily.
  • Finding and researching what the PC port added on top of that. How did they implement it?
  • Researching, finding. and understanding the windowing and rendering system on the OS variant my handheld.
  • Researching, finding, and understanding the Portmaster framework which would be used to launch games.
  • Looking into the Gamecube rendering API and comparing it to the GLES rendering pipeline, and attempting to reconcile the two.
  • Setting up a development environment to actually pull all of this together.
  • Getting things to compile and transferring it to the device.
  • Tracking down the specific cause of any errors and deeply understanding the underlying codebase to the point of being able to pinpoint and resolve the errors.
Now, in this case, it could all be done, but it would take weeks if not months of time reading through and understanding a bunch of existing documentation and code just to understand it. There isn't really any creative process to 99% of the work there. And I would personally never use any of that knowledge literally ever again in my entire life. And even if I went and did this, maybe like 100 people would ever use the result, likely for less than an hour each. So, in my opinion, it's just not worth doing.

But, you see, all those things I listed are things that AI now can do in minutes. It takes that huge process and shrinks it down to days/weeks of effort instead of months/years.

There is, ofc, a lot of credit and respect due to the folks who would go through and do it all themselves. And there is never anything to take that away from them. But if you don't have to, why would you?

Put another way - there will always be a place for artisanally-made, Alpaca wool blankets that cost $5000 each. But 95% of the time all you really need or want is a $20 throw from Costco to dump on your couch.
 
I dislike AI because it's not safe for us. AI are spywares and data thief!

Ai must die and put down in coffin. Take all AI down!

I stick with old traditional permanent without AI. Old fashion rules and always ignore AI forever. I know AI are not our friends, just enemy to us and people. AI are worst ever what we heard about these. Stop all AI then take down!

Good news, most farms rejected AI companies data center. Keep farms to produce vegetables, fruits and wheat. Never let AI companies steal farms! Stop AI companies now! We need to fight against AI companies and keep farms to build strong food supply chains to feed hungry people!

AI are full of s**t and waste their time, money and energy!

People will target nuclear power plants to destroy will take all AI down in near future. Destroy nuclear power plants to knock AI offline permanentl!

All of AI will get banned under new laws, because AI are threats to us and people! We will fight against AI right now to future!

:angry::angry:
 
I just ported a game I was working on over a few months in my spare time, to a completely different system using AI in a couple of days (also spare time). It really is becoming the world that coders envisioned back in the 1950's. "computer I want a program to do x,y,x" instead of days/weeks/months of coding its done in seconds.

AI slop is just a byproduct, just like photoshop slop before it. I didn't see all these people shouting from the rooftops that photoshop was killing the creative industry. People didn't care, but now they do, go figure.
 
Just to let you know, an overgrown Markov Chain is not artificial intelligence, it's just a huge lookup table based on existing data created by humans, most of it taken from a random pool of data stolen from people who never agreed to it in the first place, and that's not even considering leaked private data, or open source licences completely fucked over by the machine.

The day AI will be actually sentient and not reading from a huge database of existing data from which it could freely seek into then copy and paste bits that seem to fit the prompt, I will agree with you, otherwise, fuck this bubble.
 
  • Like
Reactions: Skv0ra
Just to let you know, an overgrown Markov Chain is not artificial intelligence, it's just a huge lookup table based on existing data created by humans, most of it taken from a random pool of data stolen from people who never agreed to it in the first place, and that's not even considering leaked private data, or open source licences completely fucked over by the machine.

The day AI will be actually sentient and not reading from a huge database of existing data from which it could freely seek into then copy and paste bits that seem to fit the prompt, I will agree with you, otherwise, fuck this bubble.

The term "AI" is overused, yeah. But the models are not Markov chains, and haven't been for some time.

Training sets aside (whole other conversation), the latest chain-of-thought LLM's have become useful because they have the idea of "context" - i.e., the LLM is trained on a "global" dataset, but it also has "short term memory" which influences its output. The reason LLM's have just now become useful is because their short-term memory has been expanded, and they can be pointed to Google or your codebase to "read through" some amount of data before providing your word-soup response. So, in some sense, the model has a global understanding of language and then augments that with a specific subset of data, finally generating what is essentially the "most likely" correct response.

W.r.t. coding, I am not sure how familiar your average person is with the development process, but it's essentially that. ^
 
  • Like
Reactions: ssjkakaroto
Just figured I'd post about this here, because I'm now seeing huge shifts on my team and in the rest of the industry in general, and the reality of what's happening in tech is no longer aligned with public opinion. Ofc, I'm talking about AI.

Just as a preface, a little about me is that I have a doctorate in Computer Engineering, and I have about a decade of experience working across companies including Intel, Microsoft (Xbox), and Meta (Oculus). I've also been a huge AI skeptic. Out of 40+ people in my research lab in grad school, I was the only one NOT working on AI/ML. With that out of the way...

2026 is the year of AI. It's not hype anymore. It's here. Is it perfect? No. But the absolute latest models (as in, within the last few months) are now "good enough." It is genuinely useful, mostly correct, and orders of magnitude faster than humans at tasks like research and software development.

As of the past few weeks, no one on my team writes code themselves anymore. This is also the pattern seen by any colleagues I have across the industry. If you work in tech, as a software engineer, the medium-term future of your job is piloting 3-5 AI agents in parallel like a coked-out, somehow-nerdier Shinji.

Just figured I would fire the alert flares here as the general public (and especially gamers) are still violently anti-AI. But the reality is that the genie is now out of the bottle and the tools are so good that you now are dumb if you don't use it. It makes things that were previously impossible simply due to lack of RoI completely doable. I will point to the recent Animal Crossing PC port as an example of this.

I guess what I'm trying to say is, don't immediately shit all over devs who use it (homebrew or indie, especially), because you're going to be scaring away a lot of cool projects for no reason.
For the most part I use AI as a tool, since I'm also a dev. What you have to think of, outside of our area, is people are using it as if it were their own brain, thus hampering the growth and development. That's where the AI hate comes from. For you in your position it's not really a problem, but in other areas it just floods them with slop.

Also junior positions are not really being offered anymore. How would we get more mid and senior positions without the start?

But the reality is that the genie is now out of the bottle and the tools are so good that you now are dumb if you don't use it
You have to recognize how to use it, or else you'll stay dumb as I posted above.


2026 is the year of AI
Just like the year of Linux 😎
 
it doesn't help how misinformed people are about "ai", i've had someone lecture me of my use of AI as if my pc was this giant datacenter eating up all the water in maine when i'm just locally running stable diffusion on my pc

pandora's box is open, ai isn't going anywhere, whether it be LLMs (locally hosted or datacenters), or image gen (slopGPT/flux/stable diffusion/z-image/whatever else crops up in the future)

AI is a great tool if used correctly, people complaining about it now are like the people who fought back against electronic musical instruments in the 70s
 
Last edited by Latiodile,
  • Like
Reactions: ssjkakaroto
Well, this thread is a good example of what I mean. But let us consider an example that I myself went through the other day. Once again, related to the Animal Crossing PC port:

I own an Anbernic RG40XX H device. I wanted to get the Animal Crossing PC port running on it. For those unfamiliar with development, here's what that would have looked like:

  • Finding out what the AC decompilation project actually entailed. What code was there. What wasn't. What libraries did it need?
  • Understanding the Gamecube SDK, which is probably used heavily.
  • Finding and researching what the PC port added on top of that. How did they implement it?
  • Researching, finding. and understanding the windowing and rendering system on the OS variant my handheld.
  • Researching, finding, and understanding the Portmaster framework which would be used to launch games.
  • Looking into the Gamecube rendering API and comparing it to the GLES rendering pipeline, and attempting to reconcile the two.
  • Setting up a development environment to actually pull all of this together.
  • Getting things to compile and transferring it to the device.
  • Tracking down the specific cause of any errors and deeply understanding the underlying codebase to the point of being able to pinpoint and resolve the errors.
Now, in this case, it could all be done, but it would take weeks if not months of time reading through and understanding a bunch of existing documentation and code just to understand it. There isn't really any creative process to 99% of the work there. And I would personally never use any of that knowledge literally ever again in my entire life. And even if I went and did this, maybe like 100 people would ever use the result, likely for less than an hour each. So, in my opinion, it's just not worth doing.

But, you see, all those things I listed are things that AI now can do in minutes. It takes that huge process and shrinks it down to days/weeks of effort instead of months/years.

There is, ofc, a lot of credit and respect due to the folks who would go through and do it all themselves. And there is never anything to take that away from them. But if you don't have to, why would you?

Put another way - there will always be a place for artisanally-made, Alpaca wool blankets that cost $5000 each. But 95% of the time all you really need or want is a $20 throw from Costco to dump on your couch.
AI bros after spending 5 minutes typing a prompted, then 40 minutes writing a wall of text
D3C27D52-9FDB-4AFC-8133-5DA6A47615E0.jpeg
I actually use AI as a tool for my writing and assisting to video editing. However, I feel like people defending AI and treating shit they created with a prompt as “art” or “work” really don’t understand why people have issues with AI. Equally, seem to think walls of text will somehow convince them to change their minds, walls of text that never actually address concerns, I might add. It just comes off as self-righteous shit whenever people see these walls of text.
 
I think we may see that bubble burst this year. It is simply not a business where money can be made to sustain it. If I recall, they are basically borrowing money at this point to keep it up, which means they really have no more money of their own. It's like a Gold Rush scenario. The ones digging for gold aren't really getting anything, while the folks providing the supplies are making bank. In this case, one of them is Nvidia.
 
I actually use AI as a tool for my writing and assisting to video editing. However, I feel like people defending AI and treating shit they created with a prompt as “art” or “work” really don’t understand why people have issues with AI. Equally, seem to think walls of text will somehow convince them to change their minds, walls of text that never actually address concerns, I might add. It just comes off as self-righteous shit whenever people see these walls of text.

I'm... Wait...

...

...

What?

I am so confused by this response I'm not even sure what you are saying here.

I'm not sure how any of this applies to my post you quoted, lol. I can definitely tell you're upset, though, so sorry about that.
 

Site & Scene News

Popular threads in this forum