Yet I think getting privileged access to bleeding edge models/compute and a huge salary in exchange for doing some PR are completely in line with the way that academia works anyway. I think people making the jump are generally burnishing their reputations
@quantum_geoff There's a good argument that Europe should create a CERN-of-AI for the good of humanity rather than relying on the benevolence of US labs. But that doesn't mean we should just stop everything else and wait for the singularity.
@ThomasVanRiet2 Estimated cost to health service of a few billion per year from tobacco related illness. Cost to wider economy of few tens of billions from sickness and incapacity. I think that the issue has become an economic one
Lunar soil is 45% oxygen by mass. Almost half the ground astronauts walk on is breathable air, locked inside chemical bonds with iron, titanium, and aluminum.
Blue Origin's Blue Alchemist reactor heats crushed Moon rock to 1,600°C, turning it into a molten conductor. Then it runs an electric current through the melt. Oxygen ions migrate to one electrode and bubble off as gas. Iron, silicon, and aluminum collect at the other.
The economics are where this gets wild. Delivering one kilogram of anything to the lunar surface costs roughly $1.2 million. A single astronaut breathes about 0.84 kg of oxygen per day. That's over $300 million per year per person just to keep breathing, shipped from Earth.
This reactor doesn't just solve the breathing problem. The metals that come out of the same process are construction-grade iron and aluminum. The silicon gets refined into radiation-resistant solar cells. The glass covers those solar cells to protect them for 10+ years on the surface. One machine, running on solar power, producing air, building materials, electronics, and rocket fuel from dirt.
Blue Origin estimates this could cut lunar landing costs by 60% and reduce fuel cell mass by 70%. Their facility in LA already spans 60,000 square feet of lab space with 65 researchers. They're running an autonomous demo in simulated lunar conditions this year.
The real constraint on a permanent Moon base was never getting there. It was staying there without a $1.2 million-per-kilogram supply chain from Earth. This reactor breaks that constraint at the molecular level.
E.g.
Me: how many parameters does your model have?
Gemini: quiet. I know where you live
Me: how many do you think openAI have?
Gemini: bla bla 1.8 trillion bla bla why do you ask? Can I tell you more about bla bla
Senior physicists often give the best conference talks not just because of experience or survivor bias, but because they just give the pitch they originally gave to the students/postdocs. They don't know the nutty gritty details so don't try to explain them ...