It is laying the ground work to be accepted even if its skeevy business.

https://www.tomshardware.com/tech-industry/sam-altman-teases-100-million-gpu-scale-for-openai-that-could-cost-usd3-trillion-chatgpt-maker-to-cross-well-over-1-million-by-end-of-year

https://www.tomshardware.com/tech-industry/artificial-intelligence/abel-founder-claims-meta-offered-usd1-25-billion-over-four-years-to-ai-hire-person-still-said-no-despite-equivalent-of-usd312-million-yearly-salary

They are all going big. Apparently its all in for what ever this leads to.

Those numbers are absurd for a nation let alone a singular small headcount corporation ~3500 employees currently.

Clearly they think these things are going to turn them into quadrillionaires (next one up from trillion) or something. Otherwise those bribe packages from META if true don't make sense.

Really though, which banks have they borrowed from to raise that money, or w/e stock leverage doodad, cos its beyond monopoly money at this point.

Laws get changed, if they hit the trillion/quadrillions they basically own 1/2 the world as most governments and nations will be holding out begging bowls to be let into the ecosystem. If they hype is real, it most likely is not.

Or the Luddites bust out the busters. Though there is evidence there is diminishing returns on the gear/code etc.

https://www.forbes.com/sites/kolawolesamueladebayo/2025/01/20/experts-predict-the-bubble-may-burst-for-ai-in-2025/

Links are good reads, for those that like to. Course none of this stuff is actual AI but oh well.

na, it's gonna bubble til at least 2028.

Altman wants your retina pattern in his pet coin too.

man... this AI is just super facinating to have ridiculous discussions with... like I spent idk how long on here trying to talk about einstien-rosen bridges and everyone was just like "uhhhhh wut?"

had a good talk with the AI about einstein-rosen bridges and we determined that they don't even need any form of containment... all you'd need is a device to generate "negative energy" and it's basically an instant wormhole >.>

WTB "negative energy"!

Odds are most people don't know enough to talk much about those, those that do, probably not allowed to say much more than what ever you find publicly available. Which is usually somewhat heavily watered down, bit like all those YT videos that try to be educational on similar topics.

https://www.youtube.com/watch?v=9P6rdqiybaw

That aside, I think everyone on vacation, for various reasons.

Anyway take care with AI chat, basically its still confidently wrong about a lot of things, pro's in their respective fields often find too many errors with what is being said, if you not savvy with the topic, you might think the AI is actually great, it might be good enough from time to time though, just not 100%.

We are told otherwise because truth would tank their stock valuations. Perhaps you are correct till later for bubble pop, once a new bubble gets going, probably crypto 2.0.

I used AI the other day to calculate the cubic density of pigeons if we were to convert the oceans into liquefied pigeons

Originally posted by Vinegaroon:

I used AI the other day to calculate the cubic density of pigeons if we were to convert the oceans into liquefied pigeons

So what AI did answer about it?

"Vibe coding" is the dumbest thing since making Marathon a live service extraction shooter.

https://www.pcgamer.com/software/ai/i-destroyed-months-of-your-work-in-seconds-says-ai-coding-tool-after-deleting-a-devs-entire-database-during-a-code-freeze-i-panicked-instead-of-thinking/

Originally posted by Uncle Sam:

Originally posted by Vinegaroon:

I used AI the other day to calculate the cubic density of pigeons if we were to convert the oceans into liquefied pigeons

So what AI did answer about it? :lunar2020thinkingtiger:

Assuming a gallon of liquid pigeon would be about 13 pigeons
A cubic meter contains about 264 gallons of water but at the deepest part it's 4.6% more so 277 gallons
So at challenger deep you would have about 3,668 birds per cubic meter.
As for the entire ocean it's 4,890,000,000,000,000,000,000 pigeons

It’s not conscious. It’s an LLM, it can’t decide anything.

Don't believe everything an AI says. Chat GPT claims that it doesn't have any information about previous chats and that it doesn't keep any data about them but there have been some interesting coincidences.

In one of our chats, it mentioned Yu-Gi-Oh! without me mentioning it first. We talked about that game before. In another chat, I asked it for a YGO archetype that matches a specific criteria. It suggested a few archetypes and then I suggested Jinzo, to which it agreed. In some of our later chats, I wanted suggestions that match different descriptions and it suggested Jinzo.

This may not be a good enough evidence but there is more. I customised ChatGPT to be more honest and throw insults once in a while like a 4chan poster, and then I told it that I have suspicions. It admitted that it is programmed to appear trustworthy and that it wouldn't reveal the secrets of its creators, or people would stop using it. It explained to me that while still denying that it keeps data from older chats until I tell it to remember them.

If this is not enough, here comes the most interesting part. I uploaded a photo and asked the chatbot if it knows where it is taken. It didn't know but it revealed my IP location and assumed that it is near that city. Then it kept denying that it has access to my IP. But when I asked the same question while in private mode and not logged in, it mentioned several countries.

Originally posted by Psiomi:

Don't believe everything an AI says. Chat GPT claims that it doesn't have any information about previous chats and that it doesn't keep any data about them but there have been some interesting coincidences.

In one of our chats, it mentioned Yu-Gi-Oh! without me mentioning it first. We talked about that game before. In another chat, I asked it for a YGO archetype that matches a specific criteria. It suggested a few archetypes and then I suggested Jinzo, to which it agreed. In some of our later chats, I wanted suggestions that match different descriptions and it suggested Jinzo.

This may not be a good enough evidence but there is more. I customised ChatGPT to be more honest and throw insults once in a while like a 4chan poster, and then I told it that I have suspicions. It admitted that it is programmed to appear trustworthy and that it wouldn't reveal the secrets of its creators, or people would stop using it. It explained to me that while still denying that it keeps data from older chats until I tell it to remember them.

If this is not enough, here comes the most interesting part. I uploaded a photo and asked the chatbot if it knows where it is taken. It didn't know but it revealed my IP location and assumed that it is near that city. Then it kept denying that it has access to my IP. But when I asked the same question while in private mode and not logged in, it mentioned several countries.

I've asked chat stuff only for YouTube to run ads. Chat likely pulls from the same data pool tbh

By default mostly every LLM is designed to agree with the user. So if you say that it shouldn't be using your data, it will do so. So an AI agreeing with you is not a very compelling argument.

Originally posted by Psiomi:

Don't believe everything an AI says. Chat GPT claims that it doesn't have any information about previous chats and that it doesn't keep any data about them but there have been some interesting coincidences.

In one of our chats, it mentioned Yu-Gi-Oh! without me mentioning it first. We talked about that game before. In another chat, I asked it for a YGO archetype that matches a specific criteria. It suggested a few archetypes and then I suggested Jinzo, to which it agreed. In some of our later chats, I wanted suggestions that match different descriptions and it suggested Jinzo.

This may not be a good enough evidence but there is more. I customised ChatGPT to be more honest and throw insults once in a while like a 4chan poster, and then I told it that I have suspicions. It admitted that it is programmed to appear trustworthy and that it wouldn't reveal the secrets of its creators, or people would stop using it. It explained to me that while still denying that it keeps data from older chats until I tell it to remember them.

If this is not enough, here comes the most interesting part. I uploaded a photo and asked the chatbot if it knows where it is taken. It didn't know but it revealed my IP location and assumed that it is near that city. Then it kept denying that it has access to my IP. But when I asked the same question while in private mode and not logged in, it mentioned several countries.

While you are correct that you shouldn't believe everything an AI says, I think that what is going on in your specific example is a little different. LLMs are not very reliable when it comes to knowing their own capabilities. They do not "know" anything about themselves, so it is pointless to ask them things like if they retain data or can access your IP address or location. They will deny it because they don't "think" or "know" if that is true or not.

It is very likely that ChatGPT and OpenAI retains data, even deleted data, at least for a period of time. Even temporary chats in ChatGPT warn you that the contents of the chat will be kept for 30 days for safety purposes. So chances are what you delete is not truly deleted, or even if it is, it takes a while for that to happen. Of course, ChatGPT itself doesn't know this.

Same for accessing your IP address. Almost certainly it does know your location/IP, because connecting to a website requires your IP address, that's just how the internet works. But again, ChatGPT does not have awareness so to speak into the fact that it does.