AI: The Rise of the Machines... Or Just a Lot of Overhyped Chatbots?

Sanrith Descartes

Its all the Joos' fault. Am I doing this right?
<Gold Donor>
48,968
132,449
We continue to train our replacements.
note: the robot isn't being driven, its Optimus running on its own AI.

 
  • 3Like
Reactions: 2 users

Control

Silver Baronet of the Realm
4,899
13,836
It ain't the movies. Everyone runs out of ammo in real life.
Of course, when the murderbots finally come, they won't need anything fancy like kung fu or even bullets.
I imagine something like this, only not retarded and moving faster than we can see.
robot fail GIF

Although by that point, they could probably more easily just fly by and headshot us all in our houses at speed like a RL aimbot. Plenty of room for ammo that way.

I just hope murderbots aren't the first type of bot we get to enjoy.
interesting machine GIF
 
  • 1Wow!
  • 1Weird Boner
Reactions: 1 users

Aldarion

Egg Nazi
10,908
29,768
This is a month old so may have already been posted. Where's the Shovelware? Why AI Coding Claims Don't Add Up

I haven't tried any of the "tools", but I have seen ai results to searches pop up. Really though they were just sifting through the usual garbage search results to find the meaty bits.
I can't give that enough likes. Great fucking article.

I want to quote this part pre-emptively before our local "AI expert" shows up with exactly this objection:
  1. “Well, if you just learned how to prompt properly, then you would be a 10x engineer like me.”
    Look at the data. There are no new 10xers. If there were — if the 14% of self-proclaimed AI 10xers were actually 10xers — that would more than double the worldwide output of new software. That didn’t happen. And as for you, personally, show me the 30 apps you created this year. I’m not entertaining this without receipts.
 
  • 1Worf
  • 1Like
Reactions: 1 users

Aldarion

Egg Nazi
10,908
29,768
His follow up post is good too.


It’s intermittent reinforcement (like gambling.) Occasionally it nails the prompt first try, which is genuinely impressive, mixed with many misses. You remember the jackpots and forget the hours of pumping in tokens.
I thought this comment just absolutely nailed it, why does AI "feel" so impressive but fall so short in practice? This, right here.
 
  • 4Like
Reactions: 3 users

ShakyJake

<Donor>
8,345
20,836
Has anyone checked out Claude 4.5 Sonnet's "Imagine" feature? I thought it was pretty dang amazing, and it lends to the idea that, in the future, applications will be created on the fly based on the backend data. I'm a software developer and our product owner is hinting at wanting us to update our very old WebForms application to a new UI framework to visually match another product in our offering. I'm thinking this is a bad idea (basically I don't think they realize how massive of an effort this will be) and imo both products will be obsolete in a few years, UI-speaking, as I think the way we interact with computers and applications is going to change dramatically in just a few years.
 
  • 2Like
Reactions: 1 users

Lanx

<Prior Amod>
72,295
170,529
Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th.

we're fucked when ai passes the fresh prince eats spaghetti test


 
  • 5Like
  • 1Worf
Reactions: 5 users

Deathwing

<Bronze Donator>
17,491
8,544
Has anyone checked out Claude 4.5 Sonnet's "Imagine" feature? I thought it was pretty dang amazing, and it lends to the idea that, in the future, applications will be created on the fly based on the backend data. I'm a software developer and our product owner is hinting at wanting us to update our very old WebForms application to a new UI framework to visually match another product in our offering. I'm thinking this is a bad idea (basically I don't think they realize how massive of an effort this will be) and imo both products will be obsolete in a few years, UI-speaking, as I think the way we interact with computers and applications is going to change dramatically in just a few years.
What's going to change in how we interact with computers and applications? Aside from removing/disabling/ignoring, if possible, the AI integrated into everything? I'm almost to the point where every google search is suffixed with "-ai" by muscle memory.
 

ShakyJake

<Donor>
8,345
20,836
What's going to change in how we interact with computers and applications? Aside from removing/disabling/ignoring, if possible, the AI integrated into everything? I'm almost to the point where every google search is suffixed with "-ai" by muscle memory.
I'm picturing user interfaces ditching the old desktop metaphor for something like Star Trek, where we just chat with AI, saying "show me this" or "do that," and it understands us... fast, natural, and super intuitive. No more clicking through menus.
 
  • 1Like
Reactions: 1 user

Deathwing

<Bronze Donator>
17,491
8,544
I'm picturing user interfaces ditching the old desktop metaphor for something like Star Trek, where we just chat with AI, saying "show me this" or "do that," and it understands us... fast, natural, and super intuitive. No more clicking through menus.
Have LLMs gotten cheap enough to do that? IIRC, you'd have to retrain the model every time you change the software in a way that's UI and/or documentation facing.
 

ShakyJake

<Donor>
8,345
20,836
Have LLMs gotten cheap enough to do that? IIRC, you'd have to retrain the model every time you change the software in a way that's UI and/or documentation facing.
We're years away. Not suggesting this is happening overnight.

And, Antropic's "Imagine" feature is where things are headed imo. There is no "application software". It's just data, and the AI will render whatever's necessary to fulfill the user's request.
 
Last edited:
  • 1Like
Reactions: 1 user

Control

Silver Baronet of the Realm
4,899
13,836
Is music worth looking into again? I tried it a few years ago and it didn't go well.
I've been watching this guy's channel, and he did a small review of the latest Suno recently:
I'm pretty sure he also covered an open source music generator sort of recently too, but the way he stacks up multiple topics, it's hard to search. Here's the last one I see, but it's from a year ago:
 
  • 1Like
Reactions: 1 user

Deathwing

<Bronze Donator>
17,491
8,544
We're years away. Not suggesting this is happening overnight.

And, Antropic's "Imagine" feature is where things are headed imo. There is no "application software". It's just data, and the AI will render whatever's necessary to fulfill the user's request.
How do you test this feature? AI is difficult enough to test as is, dynamically generating a UI seems too far in that direction.

I have a friend that works in AI research and to test the output of their AI, they use other AIs as baselines. o_O
 
  • 1Like
Reactions: 1 user