In this case a false denial would put them in violation of two FTC consent decrees, and would almost certainly leak (Google employees are not known for keeping their mouths shut), so it would be a particularly stupid thing to do.

The point is to explicitly extend android to be more 'laptop/desktopy'. So it's still a linux distribution, and looks like it won't be *really* distinct from Android....

They might not have had a choice. The memory vendors getting sweetheart deals from AI supply chain might require other markets to increase their commitment or get nothing.

So the choices might be either stockpile or not have any supply at all for their mainstream product. It's worth a risk of overpaying for memory when you have no other viable option.

Of course they didn't say that. They've always been open about doing that for unpaid consumer accounts, it's how they can provide the service for free. If you don't want your the ads, or for your data to be used, you can get that, starting at at $7 per month.

Do they, or could that stuff be on the surface? Pump cold water down, get hot steam back up, run it through a heat exchanger/condenser, cycle it back down again. Or maybe something other than water. You'd lose some heat to the shaft walls, but that could be acceptable.

I think all the maintenance-required parts you're talking about are where the heat is transformed into electricity, plus the safety-related monitoring of the core. With this design, it seems like all of the turbines, etc. will be at the surface, where they can be easily maintained, while the safety-related stuff just isn't an issue. Rather than designing a core that can be controlled and ramped up and down, with this system you'd designed the core to just operate at a continuous steady state for its operational lifetime until the fuel is used up, at which point you just fill in the hole.

You might make the core self-moderating so that if it gets too hot it will ramp down the fission so you don't have to worry about stoppage in the flow of water resulting in a meltdown or similar, but that would only be to reduced the likelihood of the core damaging itself before the end of its useful lifetime, not because there is any safety concern with a meltdown that occurs kilometers underground.

All of the stuff other than the weights is basically just scaffolding, though. Necessary, but the stuff that makes it "intelligent" is all in the weights.

Oh, for dialog it would suck. I'm thinking more about commanding 'sidekicks' to do certain things. Like voice command saying: "Bob, get up to that ledge (while pointing your crosshairs indicating the ledge) and provide cover with your sniper rifle". Today you can't direct non-human 'squad mates' with that level of specificity, so they do their specific scripted things or vaguely adjust their behavior in accordance to your vague command based on a press of a directional button. Natural language command of NPCs could open up possibilities to fix long-standing annoyances/limitations with NPCs trying to actively contribute to these situations.

For dialog, you lose the ability to be confident that the correct information has been conveyed to the player. So you can use it for background dialog for NPCs with no actionable info, but that dialog is going to be pretty pointless and particularly painful if it's hard to tell if an NPC is just background or has actual information for you.

How would you even use slaves in drilling a kilometers-deep borehole?

Yep, the guardrails are also improving.

Well, the LLMs don't really consist of "code" per se, but I think the AI labs are already using them to work on improving their own design. How far are they from being able to do this without human oversight and supervision? I have no idea.

In this case, it stays deep in the ground, where it came from.

The dry rock method described in the article significantly reduces this problem, because it doesn't rely on groundwater steam that has had millennia to dissolve high mineral loads. Instead, it injects low-mineral surface water into pressure-created cracks. That water does pick up some minerals from the rocks, of course, but the result is far less corrosive than natural groundwater. In addition, super-hot rocks flash all of the water to steam, and H20 in gaseous form cannot carry any dissolved minerals (this is how distillation works), so as long as they can find ways to keep the steam from pushing liquid water up into the generation equipment, there are no salts or minerals to cause problems. This is actually easier with higher temperatures, because no pockets are cool enough to avoid flashing to steam.

I'm not saying that the problem you cite isn't relevant to hot rock geothermal, but the difficulties are hugely reduced.

What sources? And when? I think the conclusions your citing predate the reasoning overlay models... and reasoning seems like exactly the right solution to both of the problems you cite.