That’s why it’s something to watch. But I do wonder if the value of first-to-market is overrated. For this to payoff, it needs to reduce payroll. I’m not sure brand loyalty would hold up if there is detectable differences in functionality.
Something interesting to watch is OpenAI and the recent consensus that they are no longer the leader on AI, with Google’s recent releases being considered superior. If they aren’t the leaders, then suddenly they aren’t nearly as valuable as they once were. It’s not just about position, but velocity and acceleration. If OpenAI can’t catch up - and I know of no reason to expect they could - then they go from being a Ferrari to a Kia. Granted, even the most delirious AI cheerleader will say there will be winners and losers, so OpenAI losing is not an indictment on AI as a whole. But there is a difference between knowing the stove is hot and burning your hand. OpenAI losing will hurt a lot of investors, fortunately just private investors at this point. It might also spook folks who saw OpenAI as the posterchild for AI investment. Should have done that IPO a few weeks ago.
No offense taken, and did skim over those points previously. Fair points. And it does illustrate a major problem I think Dems/liberals have: how much do you prioritize helping the desperate at the expense of the middle class? The desperate need it more, but neglecting the middle class loses elections.
As someone else mentioned, fiber connections-large and redundant- are critical for siting. Although, it’s less critical for an AI learning facility than most data centers. Also, this generation of chips are mostly liquid cooled because their energy density is so high. It’s literally impossible to blow enough air to cool them. Some older data centers did just have outside air for cooling in colder climates, but that’s not practical now.
Being poor is expensive. It means you can’t buy the best value in many situation. 10k in hospital bills won’t bankrupt you or me. For many Americans, 10k or 200k has the same impact on their financial security. That’s the point of insurance.
FWIW, this generation of data centers are mostly closed loop cooling with little water usage. Up to about 2 or 3 years ago, most were evaporative cooling, which are the water hogs. The big techs switched to closed loop due to community pushback. One big example was in Uruguay, which was developing a Google dc while citizens were drinking brackish water due to a drought. But closed loop is much less efficient, resulting in more power usage.
I'm curious if she has a new person in her orbit that might be influencing her. If she was exploring going for office higher than Rep, she may have brought in new/more consultants. Maybe there is one that she trusts and is amplifying. Pushing back on Trump is one thing, but she also seems to be doing it competently. That competence is surprising.
I think the problem(s) that AI would have in engineering problem solving include:
- AI is all about replicating patterns and interpolating within data. I think most of the creative solutions in engineering are either novel or pulled from a different application - breaking the pattern. AI is not going to do that. (It's the same reason why AI is shitty at writing jokes - many jokes rely on "subversion of expectations" while AI is good at matching the expectation from the pattern.)
- Most technology owners aren't going to make their stuff public. The tech companies building the data centers sure as shit aren't sharing the designs publicly. Without data to train on, there is no AI.
- AI is currently making obvious mistakes. That is much less problematic than when it makes fundamental mistakes that "look" right. It's going to get harder and harder to tell what was engineered vs. what was pattern-matched.
So, I think there is an opportunity for AI to do extremely common types of designs that are publicly available through permitting. I'm thinking residential building designs maybe. For 95% of housing, you aren't looking to innovate. Maybe road design, although it's likely I am underestimating the complexity there.
I think AI is currently a good tool for certain aspects of engineering, but it's one of many, many tools.
"reportedly through the anger he caused others..."
Is he saying one of his supporters killed Meathead, and then calls him paranoid? Jesus fucking Christ. If one of your boys murders someone, maybe consider shutting the fuck up for a while.
So I tried the heat transfer calc. Assuming 1 GW of server capacity, 40% inefficiency (so 1.4 GW of total energy to cool, I think it’s 723 acres of heat sinks. That’s assuming a perfect emissitivity of the heat sync material. I’m rusty on these calcs, but it’s a simple equation. I think I’m in the right ballpark. Just simple physics demonstrate this is nonsense. So why are they mentioning it? Maybe grasping at straws? I think it’s more likely trying to get more funding for space launch development, and throwing an AI sticker on it.
So, I was searching for some lewd joke out of this possibility, but it lead me to this question: I wonder how many Oscars have been inserted into a human orifice. I'm going to guess a good 20%. It's something to consider for next time I watch the Academy Awards.
That is fucking hysterical. Let me do a quick list of issues with that approach (without having access to the article):
- Cooling. Yeah, space is cold, but it is also vacuum insulated. The only way you lose heat is through radiation, so you would need massive heat sink panels and a cooling system that routed the heat from the servers to those heat syncs. I'd love to see the calculation on a 1 GW data center, but I am guessing acres and acres of heat sinks.
- Maintenance. You'd need to have permanent staff to maintain the data center. Ok, maybe that could reduced through robots, but sooner or later you are going to need a human to do something.
-Energy. Similar to the cooling issue, you need bring in energy. Solar panels are the obvious solution, but a shit ton of solar panels. (I think the solar panels would be smaller than the heat sinks, as you are exchanging heat with the sun. (5000K to ~300K for energy incoming vs. ~300K to ~0K for energy discharge.)
- Material transfers. Computer parts wear out. Constantly and with significant mass for a large data center. You'd need to be launching computer parts regularly just to replace drives.
Looking at the 12/4 report from the Department of Labor, it looks like initial claims are recently down, but continued claims are up.
Not seasonal adjusted initial claims were down about 20% from Nov 22 to Nov 29 and they had anticipated a seasonal adjustment of 8.6%. So, yeah, it is good news. I am curious how the government shutdown might have affected the numbers. I could see vendors who support the government furloughing employees during the shutdown.
But for continued claims, it was up 70k people from Nov 8 to Nov 15, up about 4.1% in a week. Comparing to the same time last year, continued claims are up 3.3%.
And, at this point, non-violent resistance is more effective. Killing cops, federal agents, and national guard members changes the national discussion away from the state-sponsored cruelty.
The silver lining is that her life is still going to suck. Publicly throwing cash at an idiot usually has humorous results as evidenced by countless lottery winners.
I had to check if "Crapo" was derogatory. Nope, someone actually became a Senator with that tag. Good for you, Senator Crapo. Now stop being so shitty.
Politico article on the Crapo bill. I love how it makes sure that people who can't afford healthcare in the first place are herded into plans that will bankrupt them should shit happen.
My bad. You watched a video of someone calling a black couple n*****s, and your first thought was "I certainly hope people don't overreact to this. Let's make sure we get the full story." I'm sure you think that makes you the reasonable person in this discussion, which is something else you have in common with the Cinnabon lady.