spongebobstoes

isn't this the fault of the subcontractor for agreeing to the impossible?

yes, they are. they are hugely concerned that China has loads of power, and the US is incapable of keeping up

the biggest limiting factors for AI labs scaling are silicon, then electricity. they are acutely aware of this

since 2016 ish, big tech has measured infrastructure efficiency in watts, not dollars. that's because they made multiple long term deals for power that came up short

they want to build more power of any kind, ideally renewables. but the past decade has shown time and again that they can't build enough by any means

I think your anger would be better placed with the institutional barriers to building clean energy in America that prevent these infrastructure companies from doing the right thing

transmission via HVDC is extremely difficult to build due to land permitting and right of way laws

grid interconnections take 5+ years just to get approved, let alone built

only 2 nuclear power plants have been built since 1978, each taking over 10 years

it is not only "that hard", it is impossible

1/ deploying solar takes years to go through multiple layers of environmental review at municipal, state and federal levels

2/ the generator interconnection queue takes 5+ years just to permit new grid connections

3/ building new HVDC has been blocked by permits and right of way laws, which is another barrier to effective solar deployment

4/ there are only two commercial nuclear reactors built in America since 1978, and they each took 10+ years to build

they don't have the same name, the email confusion was between Wang and Chang

the major labs want to advance science. current business use cases are a happy accident

this is not a good measure of current model capability. we need to test agents in harnesses, not models with a single prompt

test Codex, not Sol. test Claude code, not Opus

I really enjoyed this. This talk contains a powerful message of optimism, possibility, agency and power that resonates strongly with me.

Computers and technology are force multipliers for humanity. As coders, we are lucky to be gifted with such high agency and global reach, all at a fairly affordable price tag.

I love computers.

this is pretty nonsense for a small or home server. it isn't that hard to make something essentially completely bulletproof over a small surface area

the issues mainly come from sprawling enterprise infrastructure, running thousands of random endpoints across software nobody cared to write carefully

it's in the GitHub repo

chips run way hotter than 20C, and radiative cooling scales to the fourth power of temperature. check the math

radiators about the same size and weight as solar panels will do the trick

there is already a h100 in orbit

1GW of compute is a lot in 2026. comparing 100GW of annual compute to SpaceX 2026 goals does not make sense

if Starship launch cost predictions are accurate, data centers in space will happen within 10 years

token cost for the $100 video was $3.25 for Sol 5.6 and $25.05 (!) for Fable 5

for the $25 video it was $4.27 (Sol) vs $16.99 (Fable)

this seems like an interesting benchmark of a complex task, even if the output has little artistic merit. the efficiency of Sol is impressive

it's well documented that models can be adversarially trained with essentially backdoors in response to special inputs

while I am skeptical that this is happening atm, there are probably many industries where the risk does not seem worthwhile

the ability to even attempt this ambitious project is what free software is all about

did you build anything tangible with this idea?

critique of writing style isn't made better by claiming it was authored by an LLM

that's exactly what I'm saying. understanding is good

you didn't re-invent it, you learned it from a teacher and then practiced until you understood it to the required degree

it's an interesting point. is it worthwhile to struggle through an incidental task that has been solved before? we all stand on the shoulders of giants

I think in most cases, understanding is the point. we don't expect students to derive general relativity before doing astrophysics. re-invention is only a tool for understanding

when a parent answers their child's question, does it decrease the curiosity of the child?

many children have an unlimited capacity to ask "why?". many adults are the same

if the abilities of AI are finite, then we will continue to have burning curiosity, questions to ask, and discoveries to make

my comment is pushback against claiming "this generation" as uniquely doomed, and doomerism in general

light pollution already means the night sky is largely invisible outside of remote areas

I heard oai turned apple down, not the other way around

why does me using AI affect your motivation at all?

"starving artist" is not a new concept

AI doesn't stop creative people from creating. it gives creative people an additional tool with which to create

tell it "speak like a robot without affectation or emotion" in your custom instructions

it is not full duplex. I think this GPT-Live thing is the first full duplex speech to speech model

name calling detracts from your credibility instead of supporting your point

I agree that awe and accessibility are often opposed, and that children are easily inspired by things that feel tractable but not boring

I like the idea that we can teach children to feel inspiration instead of intimidation when learning how things work