AI, as good or evil as human nature allows

That won’t be an issue before the next quarterly financial reports are sent to shareholders, so it’s not an issue that a CEO is going to be concerned with

 

Maybe there's societal changes that can fix that. I mean, there has to be or else there's going to be revolution.

But companies have another issue. If they get rid of all the low level people, who's going to rise up to be middle level people in 15 years? And they can't poach from other companies because they will all have done the same.

 

But that's not the CEO's problem. He's barely thinking 15 quarters out not to mention 15 years.

 

15 months.

Just enough time to get a golden handshake and cash out his shares before the new guy dumps all the blame on him.

 

Part of the problem we have today is that that middle management layer is recruited straight from college and has no on the ground experience.

There's a glass ceiling for people without a graduate degree.

 

Ever had a boss that was a stickler for the rules? They suck!

The best bosses are those that recognize and build upon individual idiosyncrasies.

 

Chatbots Are Surprisingly Effective At Swaying Voters - The Atlantic
http://archive.today/f8Y07

 

So Fox News anchors will be replaced soon...

 

Given Knave's highlighted quote...

In fact, the most persuasive chatbots were also the least accurate.​

​

It's possible they were the first to be replaced.

 

The Nvidia cards that power AI need all the memory that can be produced, and Nvidia has the money to pay for it. That means memory chip makers are switching to providing memory for AI and moving away from the consumer markets. This does not mean that memory companies will be making new factories - they know this is a bubble and there's no use making capacity that won't be needed once it's built.

So everything that needs memory (PC's obviously, but phones and tablets and game systems) has already gotten more expensive and it going to get a lot worse for at least a couple of years.

https://www.ign.com/articles/concer...roblems-for-high-end-gaming-for-several-years

 

Well the Nvidia cards have a maximum 3 year shelf (even though the creative accountants write them off after 6 years), can't they use them second hand for consumer devices?

 

No because they are running 24/7 and with the extremely fine manufacturing process used now they wear out after that kind of abuse.

 

Whoa!

I had never heard the allegations that Sam Altman raped and abused his little sister who is 9 years younger than him since she was about 3! They are in court!

It is bad, and I really don’t suggest you delve into this. Annie Altman has repressed memories that are resurfacing. I know a few women who have had this happen, and it all sounds tragically familiar.

https://www.lesswrong.com/posts/stf...ister-claims-sam-sexually-abused-her-part-2-1

 

I asked Gemini to summarize the pro-rel in the USA thread in less than 100 words.

This massive, long-running thread serves as a "containment" discussion for the repetitive debate surrounding promotion and relegation (Pro/Rel) in US soccer.

Participants generally fall into two entrenched camps:
* Pro-Pro/Rel: Argue for sporting merit, alignment with global FIFA standards, and the potential for organic club growth.
* Pro-Status Quo: Defend the closed MLS system, citing the need for financial stability, protection of investor assets, and the fragility of lower-division infrastructure.

The thread is characterized by circular arguments and heated disagreement with no consensus.

 

You could have asked me and I would have said the exact same thing. Would’ve left a much smaller carbon footprint.

 

Yup, it is going to have a cascading effect across tech.

Supposedly OpenAI alone is hoovering up 40% of DDR5 RAM production in 2026. A single industry disrupting global markets this way should not be allowed imho. At least not in a non-emergency situation.

This isn't just going to impact consumer market laptops and desktops, but also traditional datacenters (who are also going to be at best second in line). And smartphone companies have already announced they are going to reduce the memory in their cheapest and mid-tier lines of mobile devices. So you are likely going to be paying as much as or more for a phone with less memory if you replace your device next year.

All in service of a bubble that is going to inevitably burst.

Though it depends on how long it takes for that to happen for these shortages to diminish. OpenAI has contracts for their A.I. infrastructure with companies such as Broadcom, Nvidia, Oracle, ... that are going to last for 5+ years, some of them longer. So in that sense, increasing production might still make sense, since this isn't going to be a disruption that is over in one or two years.

The question no one has ever answered is how on earth they hope to create a business model to sustain all of this. OpenAI alone is on the hook for over 1.2 trillion dollars in terms of their above commitments.

 

I want this bubble to pop so goddamned hard. I am willing to personally sustain significant financial loss just to watch every tech bro face some consequence for their deplorable behavior over the last decade.

 

They likely won’t face any more consequences than did the shitheads in the financial sector that crashed the economy in 2008.

They’ll get bailed out. Our savings and nest-eggs? They’ll be radically diminished.

 

Maybe AI can answer that!

 

With Trump/GOP in charge, that is an absolute certainty.

I’m still pissed that Obama let them off the hook. If not then, when?

 

It's time to put an end to this AI shit out of climate change reasons:
NOS News• Today, 07:01 About 45 data centers in the Netherlands use as much power as almost 1.9 million homes.
This is reported by Statistics Netherlands (CBS). Data centers are consuming more and more electricity: in 2017 it was 1.2 percent of total electricity consumption, last year it was 4.2 percent. Data centers need the electricity for the computer servers. Files can be stored on it or computer programs can run. Because the servers use so much power, they also give off a lot of heat. That is why data centers use systems that cool the rooms, which in turn consume extra power. Especially because of AI, data centers worldwide are using more and more power, says Alex de Vries-Gao. He is researching the energy consumption of technology at the Vrije Universiteit Amsterdam. "To develop artificial intelligence, a relatively large amount of computing power is needed from computers in the data centers."

Large data centers
For the figures, CBS looked at about 45 data centers that consume more than 10 gigawatt hours (GWh) of electricity each year. These large data centers together used 4720 GWh of power last year. That is more than three times as many as in 2017. The number of large data centers has remained about the same in recent years, but the electricity consumption of these centers has increased.


Of all electricity in the Netherlands over 4% goes to large (are there small ones?) datacenter.

The moment a data center is established in the Netherlands, the demand for power increases considerably. "But if the supply does not increase at the same time, households will have a higher energy bill," says Roel Dobbe. He is an AI and security researcher at TU Delft. Investments are also needed to expand the capacity of the power grid. "These are costs that are put on society." Without those investments, choices have to be made, because nothing can be added just like that, says Dobbe. "If we were to prioritize data centers, it would be at the expense of other sectors or, for example, homes."

So everything points to putting an end, or at least put those AI datacenters way at the bottom of the list of getting connected to the powergrid.

 

Yep. I thought crypto mining was bad but if this AI bubble is successful we're fcked.

 

I cannot see how it can be successful, unless it is subsidized by taxpayer money for the long term. Which can't be ruled out, unfortunately.

OpenAI is currently losing around $12 billion per quarter and they are on the hook for over 1.2 trillion dollars in their ongoing infrastructure commitments from Nvidia, Broadcom, Oracle, Microsoft, ... in the next 5-7 years. So this entire thing hinges on the emergence of a profitable business model, in the short term.