AI helps me navigate a long-ago old classroom anecdote about WWII history, economic theory, cognitive bias, and ethical norms. A strange thought experiment plus hints on how AI changes research.
Suicide attacks were never a consideration because time was on the side of US industrial power. All of the carriers used in the war were either in the slipways or in production when the war began. It was just a matter of time. The question was whether US soldiers, marines, sailors, airmen and support personnel would have the courage to stick it out until victory became inevitable. Some of this courage on an island-by-island, dogfight-by-dogfight basis might appear suicidal. But it was courage.
It was because US victory was evidentially inevitable that the Japanese turned to suicide tactics as a last hope to convince Americans to back off. That might have been a sort of courage, but it was desperation.
Interesting. The ChatGPT narrative was somewhat lengthy, and I didn't include everything. But it did offer something similar to what you say: "Suicide tactics were viewed as a sign of desperation or fanaticism, something the U.S. leadership saw in the Japanese approach but did not want to emulate," and "Proposals of this nature were usually fringe ideas, quickly dismissed by senior command as tactically unnecessary and morally repugnant."
That's a good question. It's difficult because there was never a moment when Japanese victory seemed inevitable, so we have to speculate. In popular culture, and even maybe at the time, there was some fear that Japan could pose a threat to the US mainland. But because the US signed onto the Allied Germany First strategy, that clearly was not the thinking among the high command. The Japanese were extraordinary in the war's first year, and even continue to shine here and there afterwards, particularly in cruiser and destroyer actions in the Solomons. But the question was always, how much do we divert from Germany First, not holy shit we have to do Japan First.
Japan was always secondary.
But, if you want to stipulate that we were losing to Japan, we always had the option of acquiescing to Japanese war aims, which would be just to give Japan sovereignty West of the International Date Line. We would have done that rather than engaging in suicide tactics.
I think a more relevant question is: Is the US prepared to be suicidally brave fighting Chine today West of the International Date Line?
-- If you don't mind a touch of humor, here's the most succinct presentation I've ever seen of Pearl Harbor followed by Germany First. https://www.youtube.com/watch?v=9-axJTzj0VU
-- For a more sobering fictional portrayal, there's Philip K. Dick's "The Man in the High Castle," which portrayed iron-fisted rule in eastern America by victorious Germans, followed by a much weaker, tentative rule in the West by a victorious Japan that is, nevertheless, weak and subservient to Germany. A map of Dick's America from the TV series is seen in this essay I wrote: https://graboyes.substack.com/p/life-magazines-1916-american-nightmare
-- But perhaps the most interesting map in that article of mine is Howard Burke's 1937 L.A. Examiner feature, “If We Enter a World War—and LOSE!” It show's America carved up by its enemies, but also a map titled "How Japan Could Attack U.S.: Step by Step Maneuvers by Which Nipponese Might Cross the Pacific and Swoop Out of the Air to Demolish U.S. West Coast Cities" (https://digital.library.cornell.edu/catalog/ss:3293952). My piece did a closeup from that map, with a box reading: "THE FIRST OBJECTIVE MUST BE CAPTURE OF HAWAII. THIS WOULD MEAN CRIPPLING OR ANNIHILATING THE U. S, FLEET, GIVING JAPAN ONE OF THE WORLD'S GREATEST NAVAL BASES-PEARL HARBOR."
One of my pet peeves about the early war scares, and related to interning Japanese-Americans, was the fiction that Japan could invade the West Coast. A Japanese submarine did shell an oil refinery, and a few merchant ships were torpedoed, but that was it.
The problem was logistics. It took every bit of merchant shipping they could scrounge up just to invade Malaya and the Dutch East Indies, serially; even if they had repurposed their entire southwest Pacific invasions for invading Hawaii, they did not have enough shipping to get very many troops that far, and Hawaii imports almost everything they need, including some food. Japan could not have sustained a large enough invasion force, let alone an occupation force.
They also could not have sustained their carrier air support of an invasion. Their six carriers only lost, I think, 29 planes shot down, but another hundred were shot up too badly to fly again without major repairs. That was 1/3 of their total force, and mostly from the second wave, not the first surprise wave. One or two more days would have exhausted their strike force and left them defenseless against American carriers.
What pisses me off about this is that the US military at the time knew about the logisics, yet instead of trying to calm people down, they ramped up the alarm. If the West Coast Japanese-Americans were such a dangerous threat, what about the Hawaiian Japanese Americans? All that ramped-up noise was a lie from one end to the other.
Those are some really interesting stats. As I wrote in a previous piece:
“In early December 1941, Dad had a weekend pass to visit my mother, Lois, in nearby Petersburg. (They married in 1944.) Before he could leave, he was told that all weekend passes were canceled. ‘Why?’ he asked some superior of his. “Because we’re expecting an attack from Japan at any moment” came the answer. The newspapers were filled with ongoing U.S.-Japan peace talks, and Dad was always highly aware of the news. So he asked why the camp’s brass believed war was imminent. One of the camp’s senior officers, it seemed, had a sister who was romantically involved with a Congressman who sat on the Committee on Military Affairs (today’s Armed Services Committee). Hence, the grapevine. As Pearl Harbor was under attack on Sunday morning, Dad said, ‘I was marching with a rifle, guarding a water tower in southeastern Virginia from the Japanese fleet.’ Not prone to conspiracy theories, he always cast a highly skeptical eye toward claims that the bombing of Pearl Harbor was entirely unexpected.”
“Dad was a loyal soldier, eternally grateful that the service helped lift him from the poverty that befell his family in the lead-up to the Great Depression. He said his one brush with military authorities came when he was reported for openly criticizing the internment of Japanese Americans. He had a close friend in the Japanese American community and thought internment violated basic American principles. Higher-ups told him to keep his thoughts to himself.”
Interesting oddity, the U.S. printed special currency marked “HAWAII” during the war. The idea was that Hawaii was particularly vulnerable to mass counterfeiting, which could undermine confidence in the dollar. If there were an attack on the currency, the Treasury and Fed could instantly demonetize any bills marked “HAWAII.”
Interesting about the currency. Seems like such a simple sensible thing to do, makes me want to gin up a conspiracy theory that it wasn't a government idea.
Pearl Harbor had been on alert the weekend before the attack, and I have read reports that the attacking airplanes would have been discovered, the ships would have been manned and ready, and the airfields would have been ready, had the attack occurred a week earlier.
There was also the worry about sabotage, which bedeviled Rommel in North Africa too. Do you keep all the planes on a field close together so it's easier to guard against human attacks, or do you spread them out to protect against air or artillery (ground or ship) attacks?
That was funny. As a friend and I say, WW2, the most dreadful series of events in modern history, provides endless entertainment.
While it is very interesting to spitball about Axis Victory, it does essentially become impossible once the U.S. joins the Allies, provided it stays in the war (Edith Keelor must die). I enjoyed the first few seasons of Man in the High Castle, although I have yet to read Dick's novel. There was an interesting board wargame called "Mississippi Banzai" where Japan and Germany battle over occupied America.
However, I agree with Chartertopis below that Japan lacked the logistics to invade Hawaii, let alone the West Coast of the US. We do have a tendency to sensationalize threats to the Homeland. Of course, we often get caught napping, too.
I’ve never read Philip K Dick, but I enjoyed the series a lot (with one mild disappointment with the ending). The TV series greatly expanded on the novel, with new characters, new plotlines, more action, and a more explicit treatment of the multiverse. Dick’s daughter was executive producer and saw these changes as essential for a visual medium and faithful to her father’s vision.
I'm not a self-promoter. But I did interview Charles Sweeney, who dropped the bomb on Nagasaki. It was the Japanese banzai, no surrender spirit that brought this on. Americans follow a different path. But are merciful on acquiescence.
I am confused how it is possible that AI did work that "would have required laborious journeys through physical libraries". AI is trained on digital data... everything it used was already scanned and in digital form on the Internet ... LLMs have no ability to access analog data that is not already somewhere on the Web.
I believe it’s true that search engines searched a smaller universe of data than the AI platforms. Lately, I’ve watched newly dropped streaming TV series. After watching an episode, I might go to Grok and ask, in Episode 6, Season 3 of The White Lotus, why was character A suspicious of character B. Out comes a mini-essay (usually correct, sometimes not). It’s info that comes up on no search engine that I know of. I gather that these platforms are absorbing discussion threads and other realms missed by Google, DuckDuchGo, Bing, etc. Analogously, there are documents that might be unreachable via search engines, but which are available as physical documents. I suspect that part of the story is reflected in the fact that the average electricity usage by an AI search is much greater than a search via search engine.
So in the case of Grok, it has immediate access to all of the data on X/Twitter before it's indexed by Google … so there's definitely a time lag issue there. In terms of searching a smaller universe of data, the only thing I can think of is Google Books … back when that started, Google scanned everything regardless of copyright and was sued. So in the case of so called “orphan books” that aren't in public domain but where the copyright holder died they have the scanned data but can't display it in a Web search … and I think that data may be accessible by Gemini. But if there is an archive somewhere that has documents that haven't been digitized then AI won't help you or obviate the need to go there.
In terms of American kamazaki style missions I googled “america suicide missions in WW 2” and then clicked Web (to bypass any AI summary) and seem to be getting a fair number of results… Not really seeing how AI revolutionized that Web search …
Yes, with that Google search prompt, you get a bunch of links. In general, it takes several tries, perhaps many tires, to get to the prompt that yields what I'm looking for. So far as I can tell, all the links churned up by your prompt concerned the Operation Aphrodite category of near-suicide missions. That sort of information was never difficult to find, though Google certainly was a big step forward from going to the library and hunting through the stacks.
Perhaps if I look through enough links from your prompt, I'll find something on the topic of whether America consider kamikaze-style guaranteed-death maneuvers, but I don't see any with that prompt. With your Google Search, I'd have to click through the hits, one-by-one, to determine whether they contain the sort of info I'm seeking. I'd have to copy and paste chunks of relevant text and then organize them and write a summary. All of this would consume an hour or two.
Instead, I asked ChatGPT, Grok, and Gemini a highly specific query, written in natural language without the thought processes that go into getting a meaningful Google prompt. The AI platforms are clearly drawing their info from a wider universe of data that seems to include online discussions not caught by Google. In my case, ChatGPT absorbed all this info, collated it, organized it, and wrote a coherent essay on the topic in around 2 seconds. It focused perfectly on military discussions of guaranteed-suicide tactics, which I saw nowhere in your links. ChatGPT provided links to its info sources, enabling me to fact-check its output.
But hey, I didn't need to write all of that. After doing so, I went to ChatGPT and asked, "How does a ChatGPT prompt qualify as a revolutionary advance from Google searches?" It instantly provided a 475-word summary, coherent, organized, bulleted, and ending with the following:
Revolutionary Because:
• It shifts the burden of synthesis from the human to the machine.
• It enables natural language interaction, eliminating the need to learn "searchese."
• It offers a scalable interface for human thought, merging information
retrieval, reasoning, creativity, and dialogue into one seamless tool.
This doesn’t make search engines obsolete—they still excel at live web access, sourcing exact documents, or comparison shopping—but ChatGPT represents a paradigm shift from lookup to conversation, from results to insight.
To be fair I did not spend more than a few minutes with my Google prompt since I have no idea whether the results matched what you had in mind. I guess we'll have to agree to disagree ... I grant that the essay writing ability of LLMs is a cool parlor trick but the whole thing seems like a lot of hype to me. But c'mon man ... asking ChaptGPT why it's a revolutionary advance is like asking a barber if you need a haircut ... I would bet good money that Sam Altman hard-coded the response to that question before they even got the technology working.
The classroom discussion sounds more like an instance of the principal-agent problem: how to convince someone to accept a 100% chance of death of the agent but with better benefit to the principal, rather than a 67% chance of agent death. But maybe that is just another name for decision making under risk and uncertainty. The Aphrodite mission seems to have been about as effective as sailing vessel fireships, or cavalry Forlorn Hope sorties. Thanks for the enlightenment on A.I. tools.
James Jones wrote of the American Torpedo 8 squadron, "No Japanese kamikaze ever flew a mission with greater certainty of death than the American torpedo crews at the battle of Midway." So, yes, American aviators would and sometimes did fly suicide missions.
You probably don't want to dive into such matters, but ifffff you do, "Shattered Sword" is the book to read, although it is only about Midway. It is full of tidbits about Japanese operations that fascinated me. For instance, US carriers had well-ventilated hangar bays where planes could warm up engines (a 15 minute process), leaving the flight deck clear until ready to bring the planes up for launching. Japanese carrier hangar bays were more sealed to protect against attack, so planes could only warm up once on the flight deck, which is one reason they could not launch their attack against the US carriers in time; they were constantly landing and launching their fighters in response to the constant uncoordinated American attacks from Midway Island.
But that's getting way off track, and I think I will stop right there.
Hmmm. My father was in Carlsons Raiders -- 2nd Marine Raider Battalion -- and was in the Makin Island raid. One of the survivors, of course. The raid more or less achieved its goals, but a few Marines, 6 I believe, were unfortunately left behind and later executed by the Japanese Army.
Anyway, I do not remember if he stated directly that the men were informed that the operation was dangerous to the point of being almost suicidal, but he said more than once that the men on the mission regarded it as such, but given the importance attached to it, no one tried to get out, so it was essentially a volunteer, near suicide thing.
Of course, the Raiders, and especially Carlsons Raiders, were trained in equality, cohesion, and the Gung Ho attitude.
He was also one of those who made the entire Long patrol on Guadacanal. He had some interesting stories about that. He was an ordinary farm boy for the most part.
Here is the Marine Corps version. He was a BAR man and later a Gunnery Sgt.
Another fascinating journey into history (with “two sides” as we say in the south) o
f economics and AI. Your hands on test of AI shows both its risks and benefits, something I am becoming more alert to as people push to adopt it. My real concern is whether ten years hence students will bother to use their own intelligence to enhance the results of their prompts. Just like students in the past who just did the minimum assignments and probed no deeper, this may create another wave of minds that merely mimic rather than understand. I sure hope not.
Question is not whether there will be gaps in critical thinking, but rather whether there will be MORE gaps than already exist. My wife and I have both been educators, and both of us have lost considerable faith in American education's capacity and desire to convey critical thinking skills. AI is just a tool. What we do with that tool is up to humans.
Do we think that LLM Ai will someday, perhaps soon, perhaps later, become entirely reliable? I cannot think of a reason why this outcome will not prevail. We are in the early days of Ai.
I assume that AI will never be entirely reliable. More reliable? Perhaps. On the other hand, a friend yesterday wrote to say she had just heard a commencement speaker say that "as AI gets more 'human' it makes more errors." She said that made sense to her.
I responded with this: "Did you ever seen Star Trek’s 'The Ultimate Computer'? A genius builds an AI computer modeled after an actual human brain and lets it run the Enterprise. It becomes paranoid and destroys other friendly vessels. The scientist who built it starts having a nervous breakdown and defending the computer. Ultimately, Spock asks the scientist whose brain it was based upon. The scientist, now collapsing into insanity says, “Why, mine, of course.” https://youtu.be/Q-FNPRfkOg8
> The pilots chose the tactic yielding a randomly distributed 2/3 mortality rate over an equally valuable tactic yielding a randomly distributed 1/3 mortality rate. Why, his question went, would they so choose? One admirable purpose of the exercise was to dissuade us from reflexively assuming that human beings follow behaviors mechanically predicted by textbook models.
Thinking about it, this can be rational in the sense that in the first case the 1/3 survivors become instant war heroes, whereas surviving a lottery isn't nearly as impressive.
Also from a strategic point of view Japan's kamikaze strategy was not that effective since there were no survivors of kamikaze missions to give after action feedback.
Sorry so late with this... It occurs to me that the 1/3 - 2/3 "decision" is not irrational as much as the scenario itself lacks appropriate respect for human nature. Just reading the scenario in real time, the following occurred to me: Do I really believe we can't fight our way to better odds in the not-kamikaze scenario? Who is doing the random selection, and do I trust him/them? Do I trust myself to follow through if chosen? What will happen to those who are chosen, and fail to carry out the mission? Will the reason matter, be it intentional or accidental? Do I trust others, if chosen, to carry this out? If I can't trust others, doesn't that make the kamikaze choice less "good" for new reasons, related to probability of success? I could go on for quite a bit. And note that none of these points even begin to address the cultural/moral/ethical reasons that Americans of the 1940s would be repulsed by the idea of trying to achieve a war objective in this way. Some of the pilots would have had severe religious objections to having to execute the suicide option, regardless of how it was implemented... Each of the things I HAVE cited amount to objections to the "truth" or "reality" of the scenario itself, as presented to the pilots.
The point is, on balance, I can easily see the decision to go with the non-kamikaze option with (stated) worse odds as being at least equally rational. Or, if one prefers, at least equally wise (if one admits that wisdom can arise from something other that pure rationality). I don't doubt the scenario was spun in a class or seminar, or even as a tactical discussion during the war itself, but it's hard for me to believe that men engaged in such a life-or-death endeavor would be less capable of poking holes in the scenario itself than I am, sitting in my easy chair... And I'm quite sure that it would be on the basis of poking those "holes" that the statistically "wrong" decision would be made. And at least one pilot, tiring of the entire discussion, would say, out loud, "F#^$, let's just FIGHT, guys!"
Suicide attacks were never a consideration because time was on the side of US industrial power. All of the carriers used in the war were either in the slipways or in production when the war began. It was just a matter of time. The question was whether US soldiers, marines, sailors, airmen and support personnel would have the courage to stick it out until victory became inevitable. Some of this courage on an island-by-island, dogfight-by-dogfight basis might appear suicidal. But it was courage.
It was because US victory was evidentially inevitable that the Japanese turned to suicide tactics as a last hope to convince Americans to back off. That might have been a sort of courage, but it was desperation.
Interesting. The ChatGPT narrative was somewhat lengthy, and I didn't include everything. But it did offer something similar to what you say: "Suicide tactics were viewed as a sign of desperation or fanaticism, something the U.S. leadership saw in the Japanese approach but did not want to emulate," and "Proposals of this nature were usually fringe ideas, quickly dismissed by senior command as tactically unnecessary and morally repugnant."
So, a relevant question would be whether we would have launched such attacks if a Japanese victory had seemed inevitable.
That's a good question. It's difficult because there was never a moment when Japanese victory seemed inevitable, so we have to speculate. In popular culture, and even maybe at the time, there was some fear that Japan could pose a threat to the US mainland. But because the US signed onto the Allied Germany First strategy, that clearly was not the thinking among the high command. The Japanese were extraordinary in the war's first year, and even continue to shine here and there afterwards, particularly in cruiser and destroyer actions in the Solomons. But the question was always, how much do we divert from Germany First, not holy shit we have to do Japan First.
Japan was always secondary.
But, if you want to stipulate that we were losing to Japan, we always had the option of acquiescing to Japanese war aims, which would be just to give Japan sovereignty West of the International Date Line. We would have done that rather than engaging in suicide tactics.
I think a more relevant question is: Is the US prepared to be suicidally brave fighting Chine today West of the International Date Line?
-- If you don't mind a touch of humor, here's the most succinct presentation I've ever seen of Pearl Harbor followed by Germany First. https://www.youtube.com/watch?v=9-axJTzj0VU
-- For a more sobering fictional portrayal, there's Philip K. Dick's "The Man in the High Castle," which portrayed iron-fisted rule in eastern America by victorious Germans, followed by a much weaker, tentative rule in the West by a victorious Japan that is, nevertheless, weak and subservient to Germany. A map of Dick's America from the TV series is seen in this essay I wrote: https://graboyes.substack.com/p/life-magazines-1916-american-nightmare
-- But perhaps the most interesting map in that article of mine is Howard Burke's 1937 L.A. Examiner feature, “If We Enter a World War—and LOSE!” It show's America carved up by its enemies, but also a map titled "How Japan Could Attack U.S.: Step by Step Maneuvers by Which Nipponese Might Cross the Pacific and Swoop Out of the Air to Demolish U.S. West Coast Cities" (https://digital.library.cornell.edu/catalog/ss:3293952). My piece did a closeup from that map, with a box reading: "THE FIRST OBJECTIVE MUST BE CAPTURE OF HAWAII. THIS WOULD MEAN CRIPPLING OR ANNIHILATING THE U. S, FLEET, GIVING JAPAN ONE OF THE WORLD'S GREATEST NAVAL BASES-PEARL HARBOR."
One of my pet peeves about the early war scares, and related to interning Japanese-Americans, was the fiction that Japan could invade the West Coast. A Japanese submarine did shell an oil refinery, and a few merchant ships were torpedoed, but that was it.
The problem was logistics. It took every bit of merchant shipping they could scrounge up just to invade Malaya and the Dutch East Indies, serially; even if they had repurposed their entire southwest Pacific invasions for invading Hawaii, they did not have enough shipping to get very many troops that far, and Hawaii imports almost everything they need, including some food. Japan could not have sustained a large enough invasion force, let alone an occupation force.
They also could not have sustained their carrier air support of an invasion. Their six carriers only lost, I think, 29 planes shot down, but another hundred were shot up too badly to fly again without major repairs. That was 1/3 of their total force, and mostly from the second wave, not the first surprise wave. One or two more days would have exhausted their strike force and left them defenseless against American carriers.
What pisses me off about this is that the US military at the time knew about the logisics, yet instead of trying to calm people down, they ramped up the alarm. If the West Coast Japanese-Americans were such a dangerous threat, what about the Hawaiian Japanese Americans? All that ramped-up noise was a lie from one end to the other.
Those are some really interesting stats. As I wrote in a previous piece:
“In early December 1941, Dad had a weekend pass to visit my mother, Lois, in nearby Petersburg. (They married in 1944.) Before he could leave, he was told that all weekend passes were canceled. ‘Why?’ he asked some superior of his. “Because we’re expecting an attack from Japan at any moment” came the answer. The newspapers were filled with ongoing U.S.-Japan peace talks, and Dad was always highly aware of the news. So he asked why the camp’s brass believed war was imminent. One of the camp’s senior officers, it seemed, had a sister who was romantically involved with a Congressman who sat on the Committee on Military Affairs (today’s Armed Services Committee). Hence, the grapevine. As Pearl Harbor was under attack on Sunday morning, Dad said, ‘I was marching with a rifle, guarding a water tower in southeastern Virginia from the Japanese fleet.’ Not prone to conspiracy theories, he always cast a highly skeptical eye toward claims that the bombing of Pearl Harbor was entirely unexpected.”
“Dad was a loyal soldier, eternally grateful that the service helped lift him from the poverty that befell his family in the lead-up to the Great Depression. He said his one brush with military authorities came when he was reported for openly criticizing the internment of Japanese Americans. He had a close friend in the Japanese American community and thought internment violated basic American principles. Higher-ups told him to keep his thoughts to himself.”
https://graboyes.substack.com/p/a-fathers-day-salute
Interesting oddity, the U.S. printed special currency marked “HAWAII” during the war. The idea was that Hawaii was particularly vulnerable to mass counterfeiting, which could undermine confidence in the dollar. If there were an attack on the currency, the Treasury and Fed could instantly demonetize any bills marked “HAWAII.”
Interesting about the currency. Seems like such a simple sensible thing to do, makes me want to gin up a conspiracy theory that it wasn't a government idea.
Pearl Harbor had been on alert the weekend before the attack, and I have read reports that the attacking airplanes would have been discovered, the ships would have been manned and ready, and the airfields would have been ready, had the attack occurred a week earlier.
There was also the worry about sabotage, which bedeviled Rommel in North Africa too. Do you keep all the planes on a field close together so it's easier to guard against human attacks, or do you spread them out to protect against air or artillery (ground or ship) attacks?
That was funny. As a friend and I say, WW2, the most dreadful series of events in modern history, provides endless entertainment.
While it is very interesting to spitball about Axis Victory, it does essentially become impossible once the U.S. joins the Allies, provided it stays in the war (Edith Keelor must die). I enjoyed the first few seasons of Man in the High Castle, although I have yet to read Dick's novel. There was an interesting board wargame called "Mississippi Banzai" where Japan and Germany battle over occupied America.
However, I agree with Chartertopis below that Japan lacked the logistics to invade Hawaii, let alone the West Coast of the US. We do have a tendency to sensationalize threats to the Homeland. Of course, we often get caught napping, too.
I’ve never read Philip K Dick, but I enjoyed the series a lot (with one mild disappointment with the ending). The TV series greatly expanded on the novel, with new characters, new plotlines, more action, and a more explicit treatment of the multiverse. Dick’s daughter was executive producer and saw these changes as essential for a visual medium and faithful to her father’s vision.
By the way, on my first reading, I missed the Edith Keelor reference. :)
I'm not a self-promoter. But I did interview Charles Sweeney, who dropped the bomb on Nagasaki. It was the Japanese banzai, no surrender spirit that brought this on. Americans follow a different path. But are merciful on acquiescence.
https://www.youtube.com/playlist?list=PLA86FC332B50B778F
Really interesting essay - thank you!
And thank you!
I am confused how it is possible that AI did work that "would have required laborious journeys through physical libraries". AI is trained on digital data... everything it used was already scanned and in digital form on the Internet ... LLMs have no ability to access analog data that is not already somewhere on the Web.
I believe it’s true that search engines searched a smaller universe of data than the AI platforms. Lately, I’ve watched newly dropped streaming TV series. After watching an episode, I might go to Grok and ask, in Episode 6, Season 3 of The White Lotus, why was character A suspicious of character B. Out comes a mini-essay (usually correct, sometimes not). It’s info that comes up on no search engine that I know of. I gather that these platforms are absorbing discussion threads and other realms missed by Google, DuckDuchGo, Bing, etc. Analogously, there are documents that might be unreachable via search engines, but which are available as physical documents. I suspect that part of the story is reflected in the fact that the average electricity usage by an AI search is much greater than a search via search engine.
So in the case of Grok, it has immediate access to all of the data on X/Twitter before it's indexed by Google … so there's definitely a time lag issue there. In terms of searching a smaller universe of data, the only thing I can think of is Google Books … back when that started, Google scanned everything regardless of copyright and was sued. So in the case of so called “orphan books” that aren't in public domain but where the copyright holder died they have the scanned data but can't display it in a Web search … and I think that data may be accessible by Gemini. But if there is an archive somewhere that has documents that haven't been digitized then AI won't help you or obviate the need to go there.
In terms of American kamazaki style missions I googled “america suicide missions in WW 2” and then clicked Web (to bypass any AI summary) and seem to be getting a fair number of results… Not really seeing how AI revolutionized that Web search …
Yes, with that Google search prompt, you get a bunch of links. In general, it takes several tries, perhaps many tires, to get to the prompt that yields what I'm looking for. So far as I can tell, all the links churned up by your prompt concerned the Operation Aphrodite category of near-suicide missions. That sort of information was never difficult to find, though Google certainly was a big step forward from going to the library and hunting through the stacks.
Perhaps if I look through enough links from your prompt, I'll find something on the topic of whether America consider kamikaze-style guaranteed-death maneuvers, but I don't see any with that prompt. With your Google Search, I'd have to click through the hits, one-by-one, to determine whether they contain the sort of info I'm seeking. I'd have to copy and paste chunks of relevant text and then organize them and write a summary. All of this would consume an hour or two.
Instead, I asked ChatGPT, Grok, and Gemini a highly specific query, written in natural language without the thought processes that go into getting a meaningful Google prompt. The AI platforms are clearly drawing their info from a wider universe of data that seems to include online discussions not caught by Google. In my case, ChatGPT absorbed all this info, collated it, organized it, and wrote a coherent essay on the topic in around 2 seconds. It focused perfectly on military discussions of guaranteed-suicide tactics, which I saw nowhere in your links. ChatGPT provided links to its info sources, enabling me to fact-check its output.
But hey, I didn't need to write all of that. After doing so, I went to ChatGPT and asked, "How does a ChatGPT prompt qualify as a revolutionary advance from Google searches?" It instantly provided a 475-word summary, coherent, organized, bulleted, and ending with the following:
Revolutionary Because:
• It shifts the burden of synthesis from the human to the machine.
• It enables natural language interaction, eliminating the need to learn "searchese."
• It offers a scalable interface for human thought, merging information
retrieval, reasoning, creativity, and dialogue into one seamless tool.
This doesn’t make search engines obsolete—they still excel at live web access, sourcing exact documents, or comparison shopping—but ChatGPT represents a paradigm shift from lookup to conversation, from results to insight.
To be fair I did not spend more than a few minutes with my Google prompt since I have no idea whether the results matched what you had in mind. I guess we'll have to agree to disagree ... I grant that the essay writing ability of LLMs is a cool parlor trick but the whole thing seems like a lot of hype to me. But c'mon man ... asking ChaptGPT why it's a revolutionary advance is like asking a barber if you need a haircut ... I would bet good money that Sam Altman hard-coded the response to that question before they even got the technology working.
We’ll see how it goes over the next few years. :)
The classroom discussion sounds more like an instance of the principal-agent problem: how to convince someone to accept a 100% chance of death of the agent but with better benefit to the principal, rather than a 67% chance of agent death. But maybe that is just another name for decision making under risk and uncertainty. The Aphrodite mission seems to have been about as effective as sailing vessel fireships, or cavalry Forlorn Hope sorties. Thanks for the enlightenment on A.I. tools.
Your analysis is certainly a worthy addition. Glad you enjoyed it.
James Jones wrote of the American Torpedo 8 squadron, "No Japanese kamikaze ever flew a mission with greater certainty of death than the American torpedo crews at the battle of Midway." So, yes, American aviators would and sometimes did fly suicide missions.
Interesting. Question is whether they admitted as such beforehand. Or whether they held a tiny sliver of hope of returning.
The torpedo squadrons had done much better at Coral Sea a month before. I don't think anyone expected the disaster at Midway.
That's my impression, though I'm not particularly knowledgeable on Coral Sea or Midway.
You probably don't want to dive into such matters, but ifffff you do, "Shattered Sword" is the book to read, although it is only about Midway. It is full of tidbits about Japanese operations that fascinated me. For instance, US carriers had well-ventilated hangar bays where planes could warm up engines (a 15 minute process), leaving the flight deck clear until ready to bring the planes up for launching. Japanese carrier hangar bays were more sealed to protect against attack, so planes could only warm up once on the flight deck, which is one reason they could not launch their attack against the US carriers in time; they were constantly landing and launching their fighters in response to the constant uncoordinated American attacks from Midway Island.
But that's getting way off track, and I think I will stop right there.
Hmmm. My father was in Carlsons Raiders -- 2nd Marine Raider Battalion -- and was in the Makin Island raid. One of the survivors, of course. The raid more or less achieved its goals, but a few Marines, 6 I believe, were unfortunately left behind and later executed by the Japanese Army.
Anyway, I do not remember if he stated directly that the men were informed that the operation was dangerous to the point of being almost suicidal, but he said more than once that the men on the mission regarded it as such, but given the importance attached to it, no one tried to get out, so it was essentially a volunteer, near suicide thing.
Of course, the Raiders, and especially Carlsons Raiders, were trained in equality, cohesion, and the Gung Ho attitude.
I think that fits your AI research fairly well.
Glad to get confirmation from someone with your background. Thank God for people like your father.
He was also one of those who made the entire Long patrol on Guadacanal. He had some interesting stories about that. He was an ordinary farm boy for the most part.
Here is the Marine Corps version. He was a BAR man and later a Gunnery Sgt.
https://www.marinecorpstimes.com/off-duty/military-culture/2018/02/08/the-2nd-marine-raiders-legendary-march-across-guadalcanal/
Another fascinating journey into history (with “two sides” as we say in the south) o
f economics and AI. Your hands on test of AI shows both its risks and benefits, something I am becoming more alert to as people push to adopt it. My real concern is whether ten years hence students will bother to use their own intelligence to enhance the results of their prompts. Just like students in the past who just did the minimum assignments and probed no deeper, this may create another wave of minds that merely mimic rather than understand. I sure hope not.
Question is not whether there will be gaps in critical thinking, but rather whether there will be MORE gaps than already exist. My wife and I have both been educators, and both of us have lost considerable faith in American education's capacity and desire to convey critical thinking skills. AI is just a tool. What we do with that tool is up to humans.
Do we think that LLM Ai will someday, perhaps soon, perhaps later, become entirely reliable? I cannot think of a reason why this outcome will not prevail. We are in the early days of Ai.
I assume that AI will never be entirely reliable. More reliable? Perhaps. On the other hand, a friend yesterday wrote to say she had just heard a commencement speaker say that "as AI gets more 'human' it makes more errors." She said that made sense to her.
I responded with this: "Did you ever seen Star Trek’s 'The Ultimate Computer'? A genius builds an AI computer modeled after an actual human brain and lets it run the Enterprise. It becomes paranoid and destroys other friendly vessels. The scientist who built it starts having a nervous breakdown and defending the computer. Ultimately, Spock asks the scientist whose brain it was based upon. The scientist, now collapsing into insanity says, “Why, mine, of course.” https://youtu.be/Q-FNPRfkOg8
> The pilots chose the tactic yielding a randomly distributed 2/3 mortality rate over an equally valuable tactic yielding a randomly distributed 1/3 mortality rate. Why, his question went, would they so choose? One admirable purpose of the exercise was to dissuade us from reflexively assuming that human beings follow behaviors mechanically predicted by textbook models.
Thinking about it, this can be rational in the sense that in the first case the 1/3 survivors become instant war heroes, whereas surviving a lottery isn't nearly as impressive.
Also from a strategic point of view Japan's kamikaze strategy was not that effective since there were no survivors of kamikaze missions to give after action feedback.
Both excellent points.
They often had observer planes whose only task was to watch and make reports. I don't know what percentage of attacks, other than "often".
Sorry so late with this... It occurs to me that the 1/3 - 2/3 "decision" is not irrational as much as the scenario itself lacks appropriate respect for human nature. Just reading the scenario in real time, the following occurred to me: Do I really believe we can't fight our way to better odds in the not-kamikaze scenario? Who is doing the random selection, and do I trust him/them? Do I trust myself to follow through if chosen? What will happen to those who are chosen, and fail to carry out the mission? Will the reason matter, be it intentional or accidental? Do I trust others, if chosen, to carry this out? If I can't trust others, doesn't that make the kamikaze choice less "good" for new reasons, related to probability of success? I could go on for quite a bit. And note that none of these points even begin to address the cultural/moral/ethical reasons that Americans of the 1940s would be repulsed by the idea of trying to achieve a war objective in this way. Some of the pilots would have had severe religious objections to having to execute the suicide option, regardless of how it was implemented... Each of the things I HAVE cited amount to objections to the "truth" or "reality" of the scenario itself, as presented to the pilots.
The point is, on balance, I can easily see the decision to go with the non-kamikaze option with (stated) worse odds as being at least equally rational. Or, if one prefers, at least equally wise (if one admits that wisdom can arise from something other that pure rationality). I don't doubt the scenario was spun in a class or seminar, or even as a tactical discussion during the war itself, but it's hard for me to believe that men engaged in such a life-or-death endeavor would be less capable of poking holes in the scenario itself than I am, sitting in my easy chair... And I'm quite sure that it would be on the basis of poking those "holes" that the statistically "wrong" decision would be made. And at least one pilot, tiring of the entire discussion, would say, out loud, "F#^$, let's just FIGHT, guys!"