When AI builds itself

2,072 Views | 26 Replies | Last: 1 day ago by hph6203
JB99
How long do you want to ignore this user?
AG
https://www.anthropic.com/institute/recursive-self-improvement

Anthropic is openly pushing for a coordinated slow-down on development to let governance catch up. I don't see how that can be accomplished. The genie is out of the bottle.
Logos Stick
How long do you want to ignore this user?
It can't. No idea what Anthropics angle is here, but there is an angle. There is nothing altruistic here.
ErnestEndeavor
How long do you want to ignore this user?
The same fear marketing, disingenuous, alarmist crap that Dario Amodei has been spewing out of his maniacal sociopathic mouth hole for the last 5 years.
ErnestEndeavor
How long do you want to ignore this user?
"We are slowing down because we are the good guys" sure sounds a lot better than "LLMs are no longer scaling the way we hoped and we are incapable of acheiving the rate of improvement we've been promising our investors."
Burrus86
How long do you want to ignore this user?
AG
I've seen how this movie ends…
AustinAg2K
How long do you want to ignore this user?
The government should call their bluff and take over Anthropic and shut them down because they are a public risk. These AI CEOs are so full of $#!+. If they are making a dangerous product, they could just stop making it. Imagine if a CEO in Big Pharma said, "We are developing a drug that will likely kill most of humanity. We would stop, but if we don't do it someone else will."
bmks270
How long do you want to ignore this user?
AG
Anthropic and OpenAI CEOs are both bull**** artists.

They make attention grabbing claims to keep money and hype rolling in. Their business model is what enriches investors, which is Venture investors flipping the business and selling their ownership stake to later investors.

LOYAL AG
How long do you want to ignore this user?
AG
Logos Stick said:

It can't. No idea what Anthropics angle is here, but there is an angle. There is nothing altruistic here.


The angle is trying to put in a floor for themselves to make it harder for new players to enter the market. That's what the U.S. is these days. A few companies make it to the top then pay the government to protect them from competitors via expensive regulations.
The federal government was never meant to be this powerful.
ts5641
How long do you want to ignore this user?
As stated before, AI will be the end of us.
MouthBQ98
How long do you want to ignore this user?
AG
Yhis I could buy: establishment of defensive regulations that protect their positions from emerging competition. That happens in many regulated industries. It's BS but it takes government to create and sustain an oligarchy

I don't think excessive paranoia being pushed by the left who wants to control. Information access and use and China who is desperate to catch up with US innovations is the way to go. Yes, some regulations for public safety are warranted but the level of paranoia over AI is really starting to get excessive
Loaded
How long do you want to ignore this user?
AG
Quote:


... The genie is out of the bottle.


When the 2017 paper "Attention Is All You Need" was published by Google researchers, AI as we know it today was inevitable.

If we (U.S.) don't continue to press forward, Chinese labs will take the lead and the CCP will controll the shape and values of the most powerful mind on the planet.
amercer
How long do you want to ignore this user?
AG
bmks270 said:

Anthropic and OpenAI CEOs are both bull**** artists.

They make attention grabbing claims to keep money and hype rolling in. Their business model is what enriches investors, which is Venture investors flipping the business and selling their ownership stake to later investors.




Yep.

If you tell the world that your invention is powerful enough to destroy the world, that implies that it's also powerful enough to justify the trillion dollar valuation you want from investors.

So a lot of the fear mongering is really just to make AI look really powerful as a means to raise more cash.
infinity ag
How long do you want to ignore this user?
ErnestEndeavor said:

"We are slowing down because we are the good guys" sure sounds a lot better than "LLMs are no longer scaling the way we hoped and we are incapable of acheiving the rate of improvement we've been promising our investors."

this

(never trust a CEO)
infinity ag
How long do you want to ignore this user?
ErnestEndeavor said:

The same fear marketing, disingenuous, alarmist crap that Dario Amodei has been spewing out of his maniacal sociopathic mouth hole for the last 5 years.


One more convert!
infinity ag
How long do you want to ignore this user?
AustinAg2K said:

The government should call their bluff and take over Anthropic and shut them down because they are a public risk. These AI CEOs are so full of $#!+. If they are making a dangerous product, they could just stop making it. Imagine if a CEO in Big Pharma said, "We are developing a drug that will likely kill most of humanity. We would stop, but if we don't do it someone else will."


Nothing more to add. Beautiful post. Blue star to you.
infinity ag
How long do you want to ignore this user?
Loaded said:

Quote:


... The genie is out of the bottle.


When the 2017 paper "Attention Is All You Need" was published by Google researchers, AI as we know it today was inevitable.

If we (U.S.) don't continue to press forward, Chinese labs will take the lead and the CCP will controll the shape and values of the most powerful mind on the planet.


Then we bomb China like we bombed Iran.
infinity ag
How long do you want to ignore this user?
amercer said:

bmks270 said:

Anthropic and OpenAI CEOs are both bull**** artists.

They make attention grabbing claims to keep money and hype rolling in. Their business model is what enriches investors, which is Venture investors flipping the business and selling their ownership stake to later investors.




Yep.

If you tell the world that your invention is powerful enough to destroy the world, that implies that it's also powerful enough to justify the trillion dollar valuation you want from investors.

So a lot of the fear mongering is really just to make AI look really powerful as a means to raise more cash.


You are also right. I've been saying that on here for a long time. AI has some great use cases but it won't take over the world. These doofus CEOs pump it up so that they can grift some more cash from investors. I personally don't care, I've enjoyed the ride and made piles and piles of cash for myself. These CEOs work for me!
torrid
How long do you want to ignore this user?
AG
What about when AI bills itself?
infinity ag
How long do you want to ignore this user?
torrid said:

What about when AI bills itself?


The possibilities are endless!
No Spin Ag
How long do you want to ignore this user?
Loaded said:

Quote:


... The genie is out of the bottle.


When the 2017 paper "Attention Is All You Need" was published by Google researchers, AI as we know it today was inevitable.

If we (U.S.) don't continue to press forward, Chinese labs will take the lead and the CCP will controll the shape and values of the most powerful mind on the planet.


True.

Like it or not, AI IS the new arms race and we can't afford to lose, because if we do, weapons don't even need to be launched for us to be affected in ways bombs wouldn't be able to do.
There are in fact two things, science and opinion; the former begets knowledge, the later ignorance. Hippocrates
hph6203
How long do you want to ignore this user?
AG
ErnestEndeavor said:

"We are slowing down because we are the good guys" sure sounds a lot better than "LLMs are no longer scaling the way we hoped and we are incapable of acheiving the rate of improvement we've been promising our investors."
Think it's moreso they're compute constrained, because their CEO doesn't understand the kind of financial leverage he has in terms of capital raising and debt when it comes to building infrastructure. Goober went on Dwarkesh and said they were at a $10 billion revenue run rate, heading to $100 billion the following year and if he projected a $1 trillion revenue for 2028 with an equivalent spend on compute and his revenue fell short at $800 billion he'd be bankrupt.

Like banks want a default on $1 trillion loan because growth was lower than expected. They want payments. They want your growth. They'll give you the year to catch up and you slow expansion incrementally.



He's smart, but he lacks Elon's balls. Elon is an optimistic projector, an aggressive pursuer and understands at that scale if things don't go to plan you can fix it because everyone's got pain if you don't. It's not a mortgage on a $1 million home. Why Dario is going to Elon for incremental compute at a premium.
Spaceship
How long do you want to ignore this user?
AG
We should just tell AI to slow down AI. Problem solved.
ShinerAggie
How long do you want to ignore this user?
AG
AustinAg2K said:

If they are making a dangerous product, they could just stop making it.

They know this. They won't be able to contain it. The only hope that humanity has is that the cost outstrips the benefit.
________________________________________________________
“Those who cannot change their minds cannot change anything.”
- George Bernard Shaw
Buck Turgidson
How long do you want to ignore this user?
Follow these lyrics and believe them. Priest has been warning us since 1980.

Mega Lops
How long do you want to ignore this user?
AG
Feels like an episode of Cleopatra 2525

4
How long do you want to ignore this user?
AG
ts5641 said:

As stated before, AI will be the end of us.

Ok.
hph6203
How long do you want to ignore this user?
AG
ts5641 said:

As stated before, AI will be the end of us.
Electricity is first in line for that doombringer role. AI will have to wait its turn.


Refresh
Page 1 of 1
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.