Makes me so angry. All the problems that couldve been solved with that kinda money. Climate crisis. World hunger. Population migration. Housing affordability.
If Trump triggered WW3 and we all got nuked id be fine with it. We don’t deserve to exist.
Instead, all that money is being used to accelerate our doom. AI datacenters unnecessarily consuming power and drinking water in small towns everywhere. Many just dumping humidity into the air and letting that water literally blow away via lazy evaporative cooling. Most “normal” water consuming processes consume, treat, and return water to the downstream-traveling aquifer.
Now, couple that with an overall warming climate. When air is warmer, the more moisture the air can hold. So we end up with more water vapor in the air than normal. With the weirding factor of climate change, this means more water energy for more powerful and destructive storms the likes of which humanity has never seen. Which feeds back into more ice melting, oceans rising, permafrost melting, cycle, accelerate, cycle, accelerate.
Also, real curious to see how millions of warehouses belching humidity and heat into the air across the surface of the globe can affect the general weather patterns, but that sadly won’t be known until after the damage is done.
There was an ad during the Super Bowl that succinctly sums up how I feel right now: “America deserves Pepsi”
Removed by mod
The climate crisis couldn’t be solved with such a small sum of money
852 billion isn’t ‘a small sum of money’. And thats just OpenAI - add to it what Google, MS, Meta etc have spent and I’ll bet you’ll get close.
addressing world hunger would decimate our economies
What will decimate the global economy is the correction that will happen when either AI proves to be a folly or investors realise they’ll never recoup what they’ve paid out. Or when theres mass unemployment cos AI has taken all the human jobs. We’re cooked whichever way it goes.
not taken jobs, but laying people in such massive amounts, and “rehiring” cheaper employees that have been outsourced.
I feel as though people in 3rd world countries will be hit the hardest. Most of the jobs in first world countries can already be done much, much cheaper by foreign workers. Although we could outsource everything, it feels like we give people jobs just to keep people busy.
In the next two decades, I’m convinced that the countries that can’t afford to invest in AI and automation will be hit the hardest. The value of their work will be reduced, and richer nations will not freely share the technology.
…it was my point about world hunger. Things are set up to where the global poor have just enough money to by food and shelter. Their low wages are why people in my country have such a high standard of living.
i heard indians are getting laid of by the droves as well due to the AI boom.
climate crisis doesn’t cost anything from an energy perspective. Just freedom of competition. renewable energy can outcompete incumbent energy, and a carbon tax and dividend scheme (0 cost) ensures the full (future) cost of dirty energy is considered.
I just used the voice feature in my truck to enter an address for Google maps like always, it came up as Gemini with a long speech. I repeated the address, it asked me if I wanted the location in my home city or one in a city over 400 miles away. Regression with exponential cost.
And every fake-friendly long-winded response consumes more electricity and water than it should, while also being useless.
God I hate that. “Alexa turn on sleep” was a reliable “turn on sleep scene” until “Alexa Plus” came around, and now it randomly assumes in trying to tell it goodnight and tells me to have a good night.
Same with “sixty minutes” being immediately parsed as “sixty minute timer” and now sometimes simply results in a “what about sixty minutes?”
They’ve lowered the success metrics and satisfaction a whole bunch, but don’t fret you can now hold a “conversation” with it! Complete with logical contradictions!
Bruh, if only the goddamn android auto allow keyboard input when driving. Their voice recognition is terrible for anything vaguely non English anyway. And I just typically need to type 4 letters to get the right address in
This is most unhinged take from both sides.
Time can’t exist in LLM by design: it’s just a thing that predicts next token based on previous tokens. There is no temporal relation between tokens. You can stop and resume generation at any point. How anyone expect it to “count time”? Based on what? The best you can do is add time mark to model input at some interval.
Simplifying, somewhat complex biological systems have some kind of clocks that actually chemically tick and induce some kind of signal that they can react on.
LLMs can’t do that like at all. They never will. Some other architecture that runs in cycles? Maybe. But transformer shit? Never ever.
The issue is that ChatGPT will tell you that it can do those things. Most of the hype for “AI” has been predicated on treating it like actual artificial intelligence and not the LLM parrot it truly is
I don’t think anybody is expecting an LLM to do it
what they are expecting is the product, chatGPT, to be a one-stop spot that can do basic tasks like that
Basically just rely on tool calling to a time server of some sort 🤷
deleted by creator
Why’s this need to be on the LLM? They control the app, can’t they just make a tool call out?
Hey, set a timer for 60 seconds.
ChatGPT analyzes text
You want a timer for 600 seconds, got it!
Sets timer for 600 seconds with api.
It’ll actually misinterpret “seconds” as the number 2 instead and then start a timer lengthed 60*2 which is of course 150!
Its a Large Language Model, not a Large Number Model.
How about a small timer?
Alt-Man is a size queen.
Shit like this is a reminder to me that a large portion behind some AI products’ hype are people who have no clue what these products even do. I wonder how the world would change, if these jack of all trades who
investwaste so much time into collecting ideas to fill up their pockets, instead spent more time on actually understanding the ideas they have chosen and build at least a fundamental knowledge.Upper management in a nutshell
I wonder how the world would change, if these jack of all trades who
investwaste so much time into collecting ideas to fill up their pockets, instead spent more time on actually understanding the ideas they have chosen and build at least a fundamental knowledge.I am afraid they would be even more dangerous
Sam Altman wants funding right?
Here is an idea. I would pay 1000 dollars to get in a boxing ring with this guy, and probably a lot of other people would love to get a shot at that punchable face, no?
We have solved funding.
It’s become more and more obvious that the reason he regularly looks like a rabbit caught in headlights is because he is, in fact, a fraud and not the tech genius he would like everyone to believe.
I feel the same way about Elon Musk’s stutter. I think a good fraction of the repeating sounds he does when he talks is actually him just buying time because he’s actively bullshitting. I could be wrong, but I don’t trust that MF’er.
This will actually be solved in a week. All it takes is to add the current time to each input.
Knowing what time it is now, and telling the user the timer has completed, are two separate things.
How useful is a “oh by the way your timer ended three hours ago” tacked on to the end of the next interaction?
In the video, ChatGPT was actually being used as a stopwatch instead of a countdown timer.
How about making it prompt itself when the timer ends? That solves “BTW it ended”
To be fair, timers are hard.
Lets give it a try and see how far we get:
00:00:01
00:00:02
00:00:0E
00:00:0F
The hexadecimal value of 00:00:0F relates to an almost black color with a faint of blue — many call it the Commodore black as it was the same color as the ill-fated Commodore CDTV console, which was based on the Commodore 64 home computer. It is not only a nice color — it is also a statement of love for retro Commodore consoles.
Commodore.
Commodore.
Commodore.
Commodore.
Commodore.
Commodore.
Commodore.
Commodore.
ERROR! MODEL COLLAPSE DETECTED! TERMINATING SESSION AND CLEARING CONTEXT WINDOW.
Wednesday
deleted by creator
I shall gladly pay you Tuesday for a hamburger today!
Seven. A.M. Case the restaurant, run background checks on the staff. Can the cook be trusted? If not, I gotta kill him. Dispose of the body, replace him with my own guy no later than 4:30…
Blenteenth of Farglesblat
Y2K
if you have a variable clock speed, yeah.
He looks like a guy who got intentionally hit with lacrosse sticks over and over and over in high school because he deserved to be hit with lacrosse sticks over and over and over and we should maybe make it an annual event to hit him in the face with lacrosse sticks because he looks like a guy who deserves to be hit in the face intentionally with lacrosse sticks over and over and over.
he was plucked from obscurity by PETER thiel.
Dear Scat Altman, just add a timestamp at each response that the LLM can read
Scam Altman sounds like it’s a name straight from an hltv comment section, I love it
See, it does not understand time, so in order to.Vibe code in timer funxtionaloty, they need to start feeding it clocks.
Tell me the name of person who needs an AI for setting up timers. It is useless as ****!
The irony of voice assistants like Siri and OkGoogle being more efficient before they introduced LLMs.
