I don't think a big ai lab like Deep Mind would accept leaking their IP to their competitors so that their engineers can use code completion and generation.
Look at it the other way around, there are 8 months in the year where you can turn off or reduce fossil fuel consumption.
It creates financing challenges (keep generation around for a few months in the year means more expensive power), but it also provides independence against foreign fossil fuel imports.
This is falsely assuming that attention is the primary bottleneck. It's not. If you redirected £50B into solar, wind, and battery subsidies, you would end up with straightforwardly more green energy. Solar and wind are not resource-bottlenecked technologies. You can just buy more.
Disappointingly in the West, the primary bottleneck is permitting. Both for sites and grid connections. Not quite the same as "attention", but similar.
The UK is currently building two long offshore DC-DC connections between north Scotland, where all the wind is, and the south of England, where much of the demand is. This is easier than running another line of pylons on land because fewer people will complain.
Misallocation of capital is anti-environmental. If you allocate capital to projects that have decades long delays, you have burned that capital at the start of the project, not at the end of it when you take delivery of a working power plant.
Assuming we have limited time (i.e. that there is a deadline, a “point of no return”) and money to move out of fossil fuels, it makes sense to aim for the cheapest energy source, to optimize the probability of getting there before it’s too late.
So IMO we don’t need all the expensive options, when the cheap ones are already viable.
In the same time frame you can build out enough solar, wind and battery storage that you can supply the same amount of continuous power, have money left over and not need worry about nuclear waste and a hefty teardown bill to the tune of billions of dollars.
Nuclear power is on its way out, all you need it for is deep space, marine military applications (i.e. aircraft carriers and submarines that stay submerged for months at a time) and as a byproduct of reactors used for radiotherapy.
I'm not even sure if it is worth it to keep chasing the fusion dream. Sure, as a foundation research into nuclear physics and chemistry (i.e. creating new elements), or for powering interplanetary space travel, it still makes sense, but by the time nuclear fusion can generate power... renewables will have long since taken over.
Essentially, its a game of capital and solar is accessible to (almost) every human household on the planet.
Additionally, zeroish fuel costs/operational costs, grid connection is optional, capacity to build panels is not controlled technology, raw materials are relatively cheap. So scaling production further or bringing it online is not otherwise constrained.
Nuclear can still can be strategically important as rock solid base load and for grid stability.
In the future, with better understanding and control over nuclear physics, nuclear one day might be as cheap and plentiful as batteries, but for this cycle , solar has essentially ‘won’
The two most recent reactors built in the US used a modern modular design that was supposed to reduce the chances of construction delays. They weren't late because of planning or permitting.
The modularity was supposed to enable offsite construction of modules in a controlled environment and make it more reliable. What happened in real life is that some modules needed rework and that was only discovered after delivery. That massively fucked the schedule.
That’s a pretty massive assumption at the end: you really don’t want to cut corners with safety and unless the government can simply ignore residents, very few people are willing to live near a nuclear plant so you have to anticipate delays or higher transmission costs.
While all of that’s happening, you’re guaranteed to be producing more carbon emissions before anything can be displaced.
> 10 years to the first result, and less once the planning and construction process is streamlined.
You can't streamline your way out of the need to have to build containments capable of surviving a fully laden Airbus A380 crashing into it from 43.000 ft.
I live near a NPP, these buildings are insanely tough and you need a lot of time, a lot of resources and a lot of people to build them - especially the people part is a huge constraint, construction already isn't an attractive career and the demographic bomb is hitting the trades hard.
> You can't streamline your way out of the need to have to build containments capable of surviving a fully laden Airbus A380 crashing into it from 43.000 ft.
Most of the plane is not a problem for 6ft of concrete and a steel liner. Planes are built to just about hold together. The main thing to worry about are the jet engine cores.
> Most of the plane is not a problem for 6ft of concrete and a steel liner. Planes are built to just about hold together.
It's still 575 tons of material crashing into the containment at a speed near at or even over the speed of sound. That is a lot of energy to dissipate, on top of that 250 tons of burning jet fuel is a lot of thermal energy that can quickly eat away at steel.
The NRC report states "with regard to the effects of shock associated with aircraft impact, WEC demonstrated that these shock loadings are less than those resulting from a design basis seismic event"
> burning jet fuel is a lot of thermal energy that can quickly eat away at steel
Hmmmm, it does have to get past the concrete. "for the fire aspect of AIA ... WEC demonstrated that the requirements of 10 CFR 50.150 are met."
Jenny Chase is a renewable forecaster for BNEF (Bloomber new energy finance), and the way she works around that is by looking at China's solar panel export rather than what's reported in each country.
If you're not following her, she's often sharing really interesting stuff.
For this particular case, it's probably worth it because a dermatologist visit is $200. If I were the NHS or Kaiser, I'd partner with other integrated insurance and service providers to collect as much training data as possible and fund classifier development.
reply