Personally I hate how badly internal users are served by the majority of their systems and am willing to take some calculated long-term governance risks
This, I think, is the LLM/vibe coded app’s current place to shine.
Most internal systems don’t need massive concurrency or redundancy. It’s a webapp that reduces coordination cost between 20ish people. That’s something you can typically vibe code and deploy for ten bucks a month, and create real value.
They have done. The Three Mile Island accident happened when it was being operated by Navy vets [1]. Simple training isn’t enough.
During the investigation of the accident the Admiral that built and ran the Navy nuclear program was asked how the Navy had managed to operate accident free, and what others could learn. This was his response:
> Over the years, many people have asked me how I run the Naval Reactors Program, so that they might find some benefit for their own work. I am always chagrined at the tendency of people to expect that I have a simple, easy gimmick that makes my program function. Any successful program functions as an integrated whole of many factors. Trying to select one aspect as the key one will not work. Each element depends on all the others.
So recreating that accident free operating environment requires a lot more than just training. It would require wholesale adoption of the Navy’s approach across the entire industry. Which probably doesn’t scale very well. Not to mention the Navy operates much smaller nuclear reactors compared to utility scale reactors, and has extremely easy access to lots of cooling water, which probably gives them a little more wiggle room when dealing unexpected reactor behaviour.
How many people have died on account of nuclear accidents?
Vs. coal?
Vs. not having enough energy? (eg. blackouts killing hospital ventilators, etc.)
-----
Edit: because of HN rate limits, I can't respond to a sibling comment. I'll do that here:
> Their safety record is good, but can they generate power at a cost that's commercially competitive? If it's too expensive then the plan doesn't work.
Is a purely wind/solar + battery grid viable?
Wouldn't it be better to have a rich heterogeneous mix of various power inputs that can be scaled and maintained independently?
That's almost certainly just an artifact of old data, and I typed that before realizing your URL has the year 2011 in it.
A lot more utility solar has been installed since then. And continual improvements in efficiency spread the mining related deaths over a great many more TWh.
Our World in Data covers this and every time they update the stats, solar gains on nuclear. It's currently in the lead but they haven't updated for 6 years:
>A lot more utility solar has been installed since then.
Yes, utility solar is very safe. Unfortunately rooftop solar is much more dangerous and also much, much more costly. So one has to wonder why anyone supports the massive subsidies that are still given to rooftop solar.
Note that it's massively more expensive in the US than other nations due to paperwork and regulations. But even there it costs about the same as the low end of nuclear costs per KWh.
Adding it when building the home in the first place eliminates much of the cost and danger since you don't double up on a lot of things.
Purity isn’t really important. We need to decarbonise as much of our energy grid as we can as quickly as possible since cumulative carbon emissions matter.
Does it make sense for France to replace their existing nuclear power plants with new ones? Possibly, since the existing power generation is clean so there is less rush.
Does spending the effort on building new nuclear outweigh the opportunity costs for others? Given new nuclear plants in Europe are taking 20 years to build I have strong doubts. It seems absolutely clear that wind/solar + batteries can get most countries to 80-90% clean energy faster and at lower cost. And after that happens nuclear seems a very awkward addition to the mix since it is not cost effective to run when it’s power is only needed 10-20% of the time.
(I don't disagree that a diverse mix is good, and I'm all for nuclear, I'm just saying the old "it's intermittent and can't grid form" boogeyman is no longer true. It would also really behoove Western countries to start manufacturing batteries at scale if we don't want to get a bloody nose in the future, because they're good for more than just the grid)
If it was viable it would have happened already. We have a massive oversupply of solar and wind, particularly on the west coast. Generation is the easy part.
We have terrible storage and transmission, the parts that are actually expensive.
> If it was viable it would have happened already.
It is happening, all over the world, with a persistent and rapid growth curve.
> We have terrible storage and transmission, the parts that are actually expensive.
Better cut those tariffs on cheap Chinese batteries (and aluminium for the transmission).
Not that anyone would build one in the current political reality, but China produces enough aluminium that it would be viable to make a genuinely planet-spanning 1Ω power grid connecting your midwinter nights to someone else's midsummer days.
Viability is not just 'do the physical materials exist'. Building transmission in the US is almost entirely impossible at scale because he have no political will to do so and it's a regulatory nightmare. We can't just bury an entire mountain valley under 300 feet of water or evict a county of people to make room for a project like China can.
Ignoring the hard part and saying the aluminum exists is not even wrong, it's counterproductive. Until you solve the political component the materials might as well all be sitting on pallets in a warehouse, it doesn't help any.
Indeed; my example is intended to illustrate that the expense you experience isn't entirely necessary, but rather it is in a large part simply what America* chose.
* assuming I guessed the correct continent when you wrote "the west coast".
The Three Mile Island accident is also dramatically exaggerated in the public conscience with its severity and risk factor - solely because of the default fear of Nuclear.
Oil, Gas, Coal, and random chemical plants have had much more significant accidents even in the US, but never made a blip in the public's minds.
Aren't France and Canada the ones to learn from at this point with regards to safe nuclear operation?
Would it be fair to say that because the US Navy is not running it as a for-profit power generation that would help. Like every accident seems to be a list of cost saving shortcuts being responsible
Chernobyl was supposed to be an economically viable means of generating electricity. Comparing a tiny billion-dollar submarine reactor to a power plant simply doesn't make any sense.
The reactors on aircraft carriers have a similar thermal output to many commercial power reactors. The ones on submarines are around a third of that size, about the size of SMRs like NuScale VOYGR or the Xe-100 reactors proposed to be built at Long Mott in Texas.
Chernobyl was supposed to turn low enrichment uranium into plutonium for Soviet bombs. They made design choices that compromised safety to make plutonium production more efficient.
> It’s also worth noting that the US Navy is the only organization with a perfect nuclear safety record.
But submarine/ship reactors are tiny compared with commercial reactors and 5+ times more expensive (although its hard to break out the true lifetime cost of the reactor from the submarine/ship).
Even modern commercial SMR designs (a few by companies that make Submarine reactors) are likely to cost a couple of times more per MW than large existing reactors
They're expensive because of, arguably, over regulation. The are not inherently expensive, we've just declared them so. The next response will be "all that regulation is needed" but it's arguably that the over regulation is killing people by the unintended consequences of keeping things like coal viable, etc...
You left out the culture of borderline malfeasance on the utility's part and the failure to aggressively pursue that culture on the regulator's part.
Anyway if the fine print were risk of catastrophic failure in the event of a >9.0 earthquake I think that would be acceptable (and I think a lot of people would agree with me) depending on the geography where it was to be built.
put some generators on roof. Generator placement in Jp was based on BWRs placement in US - underground, which is sane considering tornados and stuff. But it wasnt adapted there
the cost problem is lost experience to build stuff. It's not like berlins airport is so expensive because of new whistles. Many western nations forgot to build. Japanese ABWR has a lower core damage frequency vs french EPR yet EPR took 20y and ABWR - 4
They have lost nuclear submarines (USS Thresher), lost nuclear missiles, depth charges, torpedos and bombs. They have crashed nuclear ships and submarines.
Yeah, they haven’t had a nuclear reactor leak (that we know of).
I was one of probably eight people who played the Emperor: Battle for Dune RTS game, and I always think of the Fremen character sound bite whenever I see the Old Man of the Desert’s true name invoked:
POC generally means “you can demonstrate unintentional behavior”.
“Exploit” means you can gain access or do something malicious.
It’s a fine line. Author’s point is that the LLM was able to demonstrate some malfeasance, not just unintended consequence. That’s a big deal considering that actual malicious intent generally requires more knowhow than raw POC.
Specifically: the exploit extracted the admin's credentials from the database. A blind SQLI POC would simply demonstrate the existence of a timing channel based on a pathological input.
One other commenter asked a decent question - does going lighter (Zig) or harder on memory safety (Rust) confer any meaningful advantages against the phenomenon you describe?
If they can build a peaceful relationship with Taiwan without military involvement where both countries can continue to prosper we really will have a new super power
Ah, if only.
Those damn intransigent Taiwanese!
It’s almost as if they don’t want to join the PRC.
“AI is too expensive right now” is an accurate title. Plenty of things could change in the future to change that. Off the top of my head:
* end user pricing
* breakthroughs in model efficiency
* better chips to run models
Any one of these things is easily possible in the next three years. Probably sooner.
reply