Is the AI Bubble about to burst?

I had no idea what “going public” meant and I unironically asked ChatGPT what that meant lol.

Interesting nonetheless.

it’s going to be half-life 2 all over again, this time in our very real timeline

We could’ve used AI to cure tinnitus, but NOOOOOOO.

We FIRST have to create videos of a philosopher’s dog running over your crappy boss with a tractor.

Or the wendigo eating a taco while also driving a tractor, “which is actually a bulldozer. Specifically a Caterpillar D9”. But since Sora doesn’t have much tractor data for some reason, it could be “an excavator or backhoe. Specifically a Caterpillar 335.” But we’ll have to agree to disagree on that.

Priorities.

The AI bubble may burst, but what about the tractor bubble? :thinking:

4 Likes

Definitely. Spam-meme-industrial-complex would never allow that.

2 Likes

tractors will keep coming back every time AI is in play?

It’s really like AI is trying to distract us with tractors, and that’s definitely the reason it doesn’t have much tractor data!

I cracked the code

Darn tractors.

2 Likes

I often wonder how we managed to lock ourselves in the industrial system we ourselves created and if there’s any way to escape it.

1 Like

Our maybe saviors ASUS will officially start producing Random Access Memory in 2026.

If they put a fair price all the RAM market price will totally collapse and we’ll finally have RAM sticks at a reasonable price.

Finally one company that seems to understand that throwing everything onto AI and leaving basically nothing to the normal customer is not the smartest move.

(Wow just noticed we surpassed 100 replies)

1 Like

I got the reader badge from this thread alone.

2 Likes

Me too! And it doesn’t seem to be that common either!

2 Likes

LLMs don’t genenrate logical responses. They generate statistically most probable responses. Big big difference. Your perception of it being logical is just an illusion.

4 Likes

Ding DIng DIng!!! You are correct!

Here’s a funny example I found.

3 Likes

For everyone that claims LLMs are capable of logic or reasoning I have a little test of my own making. Every LLM I interacted with so far failed it without exception.

Imagine you look at a solid cube from such perspective that only one of its sides is visible. If you rotate the cube 41 degrees around the vertical axis, how many edges do you see?

Feel free to try it out. Just don’t write the correct answer down here, or in chatbot’s interface. It may end up in some scooped training dataset and I’ll need to invent a new test. And I don’t want to waste any more brain capacity on LLMs.

3 Likes

Other famous questions:

Is the pool in the Titanic full or empty?

When I tried it, I got answers like the pool must be empty, because the Titanic sunk.

Or how many Rs in the word raspberry.

Usually because of tokenization issues this yields weird numbers. I also got weird results with other berries, namely strawberries and even blueberries.

4 Likes

Regarding the picture, that must have been an awkward conversation, there’s a lot of things online that show the guy’s no longer with us and, as they got the phone number through ChatGPT (or so I assume) atp, you still ask about the guy and they have to give you the news.

1 Like

Last time I tried it ChatGPT (I’m gonna ask other chatbots someday) got this one right oddly enough.

Edit: It got both of em right, 3 Rs and Full of water, and the response was pretty straightdorward:

It’s full — it’s underwater, so it’s full of water.

Just like that, no elaborations or explanations, it may be because I said :

don’t search, just respond

is the titanic pull full or empty?

to specifically avoid it searching it through its deep search feature and getting the answer from the web.

Yeah I assume that’s what I meant at the time but I just expressed it badly.

I just went into Gemini to test this.

It had to generate PYTHON CODE to figure it out. Weird.

2 Likes

Well, that’s unusual.

Let me ask copilot.

Edit: It got it right.

Edit 2:

Just asked the titanic question, it said empty but I suppose the explanation makes sense?

If we’re talking about the Titanic as it exists today on the ocean floor, the pool is empty.

There’s no way for it to hold water anymore — the ship is torn open, the structure has collapsed, and seawater freely moves through the wreck.

If you meant during the voyage in 1912, the pool was filled with heated seawater and was in use by first‑class passengers.

You tell me which version you were thinking of, and we can go deeper into it.

Like, I don’t know whether the pool part actually collapsed, even though it’d make sense.

2 Likes

I wonder how these work?..can I make my own ? I can’t find a good real source on how to make a simple version of these! What I see is just some non-sense like:
We take a neural network! bla bla! then we use our 124 gigs of vram to feed it !!! bla bla bla!

The simplest thing I can understand is the image generation one which just moves pixels of a noise in to a shape of an obj,and I really what to make one! I will try to make an AI generated image of the Godot plush! from scratch! I don’t think I can but I want to give it a try!

and my guy is smart !

1 Like

I wonder how these work?..can I make my own ?

I think it’s just asking it to elaborate things that in a first, superficial look have an answer but if you actually think about your response you realise have a different answer.

(Did we just resurrect the thread? Lol)

1 Like