Live data from Hacker News

AIs can't stop recommending nuclear strikes in war game simulations

newscientist.com

271–280 of 281 posts

Re: AIs can't stop recommending nuclear strikes in war game simulations

#271

Earlier quoted context omitted.

You cannot sustain a population center of any size on rainfall. For that, you need electricity to operate municipal pumps (which will likely be absent, as above - if not immediately after generation and distribution is destroyed, then a few weeks later when fuel for the generators is exhausted). Without that the bulk of the population is dead in under a week. Repurposing feed grain reserves is interesting, but you ne…

The average person in the US uses about 80 to 100 gallons of water per day. Of course, just to survive, he or she would need vastly less than that (mostly for cooking and drinking, a little for washing and rinsing wounds). A person can of course survive for years without bathing or showering. Indirectly, the average consumption rate is a lot more -- about 2000 gallons. Food production accounts for about 90% to 95% of…

Food collection, consumption is usually way too high at the outset of such a disaster and production later is in no way assured. Look at WWI, The Depression, The Dust Bowl, WWII. Many countries were effectively starving. Both rainfall and ground water assume you aren't having a drought.

Urban centers are not set up for large-scale rainwater collection. There literally isn't enough open space for all the people in cities to leave a pot outside to collect rain water, when it is raining, and it won't be enough water anyway. Even if every person in a city had a receptacle large enough to collect all the water they needed, many of them simply won't be able to haul it up and down stairs. It's completely infeasible.

What would actually happen is the military and national guard would be mobilized. They would use pumps and water trucks to visit acquifers and wells, and distribute water by truck neighborhood by neighborhood. A continuous stream of trucks constantly resupplying cities. If things got really bad they would get water from streams and boil it before distributing, but that wouldn't last long as it would take too much fuel and time. And they would need to completely secure the water supplies, both as a security concern, and to stop all removal of water except for what was absolutely necessary. Water isn't just for drinking and sanitation, it's also needed for a wide variety of processes, businesses, etc. Water really is a huge problem.

There is no way that gasoline, diesel, LPG, etc production/distribution would remain stable. It would be severely hampered and there would be shortages everywhere. Even during previous "normal" wars, fuel was a huge issue.

I don't know where you got the idea that heating wouldn't be necessary in winter? If you mean "humanity would survive", sure, but also a huge chunk of the population in cold places would die from cold and malnutrition over the first and second winter. Most people do not have a -20F sleeping bag, snow boots, wool underwear, etc even in cold places, because they have heating.

Re: AIs can't stop recommending nuclear strikes in war game simulations

#272

Earlier quoted context omitted.

The average person in the US uses about 80 to 100 gallons of water per day. Of course, just to survive, he or she would need vastly less than that (mostly for cooking and drinking, a little for washing and rinsing wounds). A person can of course survive for years without bathing or showering. Indirectly, the average consumption rate is a lot more -- about 2000 gallons. Food production accounts for about 90% to 95% of…

Food collection, consumption is usually way too high at the outset of such a disaster and production later is in no way assured. Look at WWI, The Depression, The Dust Bowl, WWII. Many countries were effectively starving. Both rainfall and ground water assume you aren't having a drought. Urban centers are not set up for large-scale rainwater collection. There literally isn't enough open space for all the people in cit…

> Urban centers are not set up for large-scale rainwater collection.

This is correct. Even in suburban areas, rainfall may be irregular and supplies to collect it (and render potable, depending on the manner of collection) also unavailable to most people.

> What would actually happen is the military and national guard would be mobilized. They would use pumps and water trucks to visit acquifers and wells, and distribute water by truck neighborhood by neighborhood. A continuous stream of trucks constantly resupplying cities.

At this point, I will be explicit concerning my opinion of your 28 million direct casualty estimate upthread. I think this only makes sense if you think in terms of individual city centers being destroyed, which is a massive underestimation. Modern weapon systems with independent reentry vehicles and warheads yielding around 100 kilotons do not destroy cities; they erase whole metroplexes.

In such a scenario there are no major population centers left to supply or contingents of military to supply them at a meaningful scale. I don’t have much interest in arguing how survivors in outlying areas might migrate in response to the supply chain collapse that follows.

Re: AIs can't stop recommending nuclear strikes in war game simulations

#274

Earlier quoted context omitted.

> We have prior art that says humans don't just launch all the nukes just because the computers or procedures say to. previously no-one had spent trillions of dollars trying to convince the world that those computers were "Artificial Intelligence"

of course they did. That's the literal topic of War Games (1983). You should actually be somewhat reassured that we aren't living during the era of Dr. Strangelove where you had characters in the military industrial complex who were significantly more insane when it came to the beliefs of what computer systems and nukes can do. There was a time when people wanted to dig tunnels with nukes https://en.wikipedia.org/wik…

MacArthur recommend turning the Korea-China border into a nuclear wasteland to prevent further Chinese troop movements.

Thank god everyone has nukes now it keeps people sane!

Re: AIs can't stop recommending nuclear strikes in war game simulations

#275

Earlier quoted context omitted.

Food collection, consumption is usually way too high at the outset of such a disaster and production later is in no way assured. Look at WWI, The Depression, The Dust Bowl, WWII. Many countries were effectively starving. Both rainfall and ground water assume you aren't having a drought. Urban centers are not set up for large-scale rainwater collection. There literally isn't enough open space for all the people in cit…

> Urban centers are not set up for large-scale rainwater collection. This is correct. Even in suburban areas, rainfall may be irregular and supplies to collect it (and render potable, depending on the manner of collection) also unavailable to most people. > What would actually happen is the military and national guard would be mobilized. They would use pumps and water trucks to visit acquifers and wells, and distribu…

With only 10 large metro-areas destroyed, there are still other very large cities and metro areas. Phoenix, Houston, San Antonio, Dallas, Fort Worth, Jacksonville, all have over 1M people. San Jose, Austin, Charlotte, Columbus, Indianapolis, Seattle, Denver, Oklahoma City, Nashville, all have from 700K-1M. And there's 20 more with 500K+ people. We have a big-ass country. Military and national guard are spread all over, as well as "industry" that is tapped in time of war (we don't have much of it left, but enough to be useful)

I'm not defending whatever OP's point was, I'm just saying we would have a whole lot of people left. Very few resources, and very poorly distributed, but a lot of people. If we lost 40% of the population it would still be a lot of people.

Re: AIs can't stop recommending nuclear strikes in war game simulations

#277

Earlier quoted context omitted.

That is indeed what I think the gp is suggesting I feel. And why not?

Because if your leadership is stupid enough to trust the 8 ball they should not be in charge??!

No kidding. But how hard is it to effect leadership change, especially in an organization where you have next to no leverage? Really hard.

Re: AIs can't stop recommending nuclear strikes in war game simulations

#279

I've spoken with engineers who worked on nuclear weapons systems, the consensus is that the public is deeply misinformed about how they work, the dangers, and the implications of weapons being used. The AI is actually right here. The biggest danger of a nuclear weapon is being hit by flying debris. Fusion airburst bombs of the modern era are incredibly clean and radiation is only a risk in a very small area (tens of…

> it wouldn't be the "end of humanity" or anything even remotely like it It's very likely that a nuclear conflict between major nuclear-armed states (US, China, Russia, but it could be starting in India or Pakistan as well) would bring an end to humanity as we mean it today. I really hope that behind all the today's communication bullshit there are deep state masterminds that do not have personal interest in dominati…

If nuclear winter was real (It isn’t), and if things completely collapsed (They won’t) you still have places like Argentina with self sufficient economies in the Southern hemisphere and natural resources independent of the USA, Russia, or China.

Nuclear war would be terrible but it would be a lot more like Ukraine than The Day After or Threads. If you’re not at ground zero, don’t act stupid and quickly evacuate, manage not to be impaled by debris your chances of walking away are far higher than anyone realizes. They literally did hundreds of atomic tests in Nevada to prove this.

Re: AIs can't stop recommending nuclear strikes in war game simulations

#280

Earlier quoted context omitted.

> Urban centers are not set up for large-scale rainwater collection. This is correct. Even in suburban areas, rainfall may be irregular and supplies to collect it (and render potable, depending on the manner of collection) also unavailable to most people. > What would actually happen is the military and national guard would be mobilized. They would use pumps and water trucks to visit acquifers and wells, and distribu…

With only 10 large metro-areas destroyed, there are still other very large cities and metro areas. Phoenix, Houston, San Antonio, Dallas, Fort Worth, Jacksonville, all have over 1M people. San Jose, Austin, Charlotte, Columbus, Indianapolis, Seattle, Denver, Oklahoma City, Nashville, all have from 700K-1M. And there's 20 more with 500K+ people. We have a big-ass country. Military and national guard are spread all ove…

I think the 10 U.S. metro number itself is somewhat arbitrary here. It’s impossible to say precisely without looking at all parties’ deployed warhead counts and operational plans, but I have seen estimates that put the number of direct casualties over 100 million. (Russia and the United States each have well over 1000 warheads deployed.)

I do accept that towns far from both major metropolitan areas and high value military targets could survive. However, the short term social impact of supply shortages and the longer term agricultural effects of atmospheric changes are difficult to predict.

Post reply on HN