Heads up: I left a Peerweb link open overnight (~8 hours) hoping it would load. It consumed ~15GB of memory and triggered an OOM kill, crashing the browser. Browser: LibreWolf (latest) OS: Fedora 43
I also had success, less negotiating, more just helping me form the letters in a few minutes rather than hours so I could get reimbursed for denied coverage, and it did get reimbursed.
Devil's advocate would say: They could do this and make it look like a bug that never gets fixed in order to avoid backlash. How it gets achieved is flexible if the goal is met.
Why would they be afraid of backlash on such an obscure, technical feature? They never were in the past and are expected to take controversial technical decisions by now. And by “now”, I mean in the last 30-odd years.
Actually I would go with 70 unless people working in industry (CEOs) is forced to retire at 60.
CEOs of large companies really run the US, people working for them tend to write the laws plus they really decide who gets to run via contributions. Most people would call them contributions, but Citizens United changed that.
Sure, we may be heading right into a massive climate crisis in the next few decades, but what about the hypothetical post-humans who will dig up spent nuclear fuel 10_000 years from now?!
Transportation and storage are still issues we have to deal with today, though.
There was definitely a window, maybe fifty years ago, where widespread adoption of nuclear energy would have stopped and reversed climate change. We may have had an increase in Chernobyls with the proliferation of non-modern reactor designs, but in this hypothetical reality maybe people would have been OK with that.
The problem today is that renewables are getting too cheap and too good, and the storage problem shrinks every day. Meanwhile, it takes upwards of a decade to license and build a single reactor. France's fancy new reactors won't be online until 2040. Nuclear is just too slow.
I feel like 10 years from now it won't even be a debate or a contest, nuclear will just be the most expensive option by a country mile. Greenhouse gases will peak within the next 2 years[0]. The nuclear lobby missed their chance, which does mean we'll have to deal with some effects of climate change we could have avoided.
My feeling right now (and I kinda flip-flop every few years) is that nuclear lost and it's economically infeasible to try again. Change my mind?
Is inside an abandoned salt mine really the same as "under the carpet"? I don't see what scenario people have in mind when they're worried about nuclear waste storage.
Do you realize it can be reprocessed yes? It's not like we don't know how to deal with it, there are both working reprocessing plants and breeding reactors for it. Only about 5% of it is actual waste and it's dangerous for much less that 10k years and can be solidified if needed...
Amazon already have a datacenter with an attached nuclear power plant, and Microsoft have contracted with Helion for future delivery of a fusion power plant (assuming it materializes).
Zuckerberg says part of the problem is not just the amount of power needed for a SOTA AI-training datacenter (~1GW), but the fact that you need the power at that particular location which makes co-locating a power plant the best option. The biggest solar power plants in China put out over 2GW, but the biggest in the US is Solar Star in CA, wihch occupies 12 km^2 and only puts out 58MW.
Thank god for comments like this. I find it truly disturbing how much people are treating LLMs like fact-machines. They are pattern matching machines that match using whatever information they have that matches best.
It's closer to improv jazz than to factual authority; still super wonderful, and worthwhile listening to, but not really for the purpose of learning how the original sounded when it was first recorded. Sure, you might get a sense of the original, but that's all it is: an impression. When you ask for facts, you get an impressionist render of facts, which sometimes, maybe even often times, accurately depict the Truth. But sometimes they depict the Truth the way an artist depicts the truth: if it feels right, it's right.
Which indicates (correctly or not) a current capacity of 314MW for Solar Star 1 (still less than the 1GW needed for Zuckerberg's projected SOTA data center).
Google search for "largest solar farms" or similar, but my bad - should have checked the date of the source.
The point I was trying to make, echoing Zuckerberg (who noted that power, not chips or data, is the constraining factor for further LLM scaling), is that power needs to be near to the data center, else the lead time and red tape will be even longer. If we're considering clean power then solar is an option, which limits datacenter location to where that is viable on this 1GW scale.
It's funny that you're the second person in this thread to assume that just because the data was wrong (out of date as it happens) it must have come from an LLM (which it didn't). I guess this is the world we are moving into, where all content is suspect of being AI generated and therefore suspect!
I was curious how Helion were coming along and found quite a good youtube "Will the Helion Fusion Reactor ACTUALLY work? - Nuclear Engineer Reacts to Real Engineering." The answer seems to be maybe eventually but not for quite a while. (vid https://youtu.be/lb7GXi0ZvYw)
The huge solar power plants in China are in western China, aka deserts and frozen desert mountains, aka you can't build data centers there either. And shipping that power across thousands of KMs is as hard for China as it is for the US.
Hence China is still building coal plants, and won't really stop. Especially if it means the future of AI is being bottlenecked by it.
PRC is attempting to optimize large compute clusters near renewables via "East Data West Computing" / "Eastern Data, Western Computing". Peak coal is expected in the next few years, new "cleaner" plants mainly replacing old plants. But there's is very much ongoing project to colocate computing/data closer to renewables. If choice is between coal + bottleneck AI, they'd chose coal. But with renewable projected to be 15-30% cheaper than coal in coming years, especially with PRC indigenous semi nodes behind (more power ineffecient per unit of compute), they'd go with renewables + shuttling data around.
"Discovery" fits since we're just uncovering what's already there, not creating it. "Invention" is for stuff we actually make. Unsure if just a mistake or creationist perspective in the article.
