The male/female ratio in China isn't that skewed, it's something like 106:100. A lot of it is probably the same thing you see in every other country, where both men and women lose social skills and have unrealistic standards of what kind of partner they are entitled to.

Right-wingers have been loudly cheering the inevitable economic downfall of California for years as its economy continues to grow. It must be so infuriating to them that Commiefornia just keeps getting richer.

"After 9 years as a US citizen, he could run for the US Senate. He would be 76 (or 74) years old.In short, it looks highly unlikely that he could (or even would) run for the US Senate."

Yeah, he would be way too young for our Senate.

I feel bad for entry-level folks -- though continually surprised how many of them outsource so much of their thinking to AI themselves. It seems like most college students cheat with AI, making them less useful as entry-level employees.

And you think the fact that carbon dioxide has steadily increased over the past 150 years or as we pump a lot of carbon dioxide into the atmosphere is a coincidence? You honestly believe that there is zero connection between pumping gigatons of carbon dioxide into the atmosphere, and there being gigatons more carbon dioxide in the air than pre-industrial times? Completely unrelated events?

Just a couple of random notes about the practicality of this.

• Some new inference data centers don't have gens. The idea is that with inference if a resource is not available it can be requested from another pool (availability zone). In these cases when the power goes out, the UPSs hold the load for 4-5 minutes to give the servers time to drop and then the site goes dark.
• Imagine 32 diesel freight train engines all powering up. That's what this is like for a site-wide utility power failure for a smallish data center. Thankfully most DCs now require an enclosure for each gen which limits the noise they produce while running.
• Most providers have between 24 and 48 hours of diesel fuel on-site. This is either in large central storage tanks or in belly tanks dedicated to each gen. A provider would have typically a 4-hour SLA on fuel delivery so once the gens start running a call is made to start running fuel trucks to the DC to begin the refueling process.

>On the bright side? Maybe watermarks can help AI stop training on AI-generated text.
 
You're living in 2023 if you think any company worth their salt is still scraping the live internet for quality content. Everyone has their own curated library already Nobody has been scraping en masse for at least two years, unless there's a blog article about bleeding edge science research or something. Even then they will probably just scrape the scientific paper itself, rather than coverage of it.

You enjoy your horse dewormer, I'll take the mRNA gene-editing.

I've been on google fi for... I dunno, probably at least a decade now? Maybe more. They just mail me a phone, I turn it on, and it figures everything out. I get a new phone ever 2 years or so. I couldn't tell you what the inside of a phone store looks like because I've never needed to go into one.

It's probably a way to prevent model collapse. So they don't train LLMs on data that has been generated by LLMs.

Clearly you know nothing about the spiritual satisfaction of using a hand-whittled oak NVIDIA Blackwell Ultra B300 rather some soulless chip made in a factory.

>the AI bubble is going to burst and
 
Kimi K3 docs and tooling, and weights, dropped this morning. Kimi is selling K3 tokens at $3/million tokens and if you're running it using GB300 gpu racks @ $6 million each... the math mostly checks out. Anthropic is selling what some people call roughly equivalent Fable tokens at $50/million each. I don't think AI is going anywhere, but valuation of companies like Anthropic, OpenAI in the $500b-$1T range isn't sustainable when "open" models sell tokens at a profit at $3/million tokens.
 
Open Weight models have been catching up to SOTA for a while but the lead OpenAI and Anthropic have, has been eroding quite rapidly and their ability to charge a premium for their product may not continue indefinitely at this point.
 
Again I'm not saying AI isn't useful, just that the bubble is largely based on valuation and Kimi K3 is.... well positioned to pop that bubble.

we're about to buy a $2400 m4 pro mac mini to grind away at a useful but low-medium priority task in perpetuity using one of the qwen 3.6 models. The ROI vs haiku api cost is less than 5 months. SOTA models are cool but a quantized 30B class model can do most any batched work if structured carefully, and you have a fairly flexible failure budget

The big issue is GDDR6 memory which is pretty new and there's not a ton of capacity already, and it's all been bought out through the end of 2028. Plants take 3 years to come online and are expensive (Read: risky) to build if the demand falls off
 
I would love to be able to buy a 4TB GDDR7 machine and never have to use cloud services again, but it's not happening before 2035

17 more years? I can't wait to see people in their 70's through 90's adventuring through the universe. Are stargates handicap-accessible?

Seriously though dude this is actually interesting science news vastly more important than some cheesy show.