Pearl Harbor — How AI Could Have Predicted and Prevented the Attack
December 7th, 1941….a day that would live in infamy. These were the words of President Franklin D. Roosevelt following the devastating sneak attack on Pearl Harbor by the Navy of the Japanese Empire. However, while many called it a sneak attack at the time and in many schools it is still taught as such, it was an attack that should have been anticipated and prevented.
There were three decades worth of warnings prior to the attack. So how did we fail, to use a 9–11 term, to connect the dots? The answer lies with human biases. Throughout history, enemy intentions were never hard to predict. After all, our adversaries are particularly vocal about their intentions. The United States, like many great powers, often gets surprised because they dismiss seemingly weaker adversaries as not having the capability to to any serious damage or we assume they follow the same decision making process and values as we do. We assume they won’t attack because we, in that position, would not do it.
So let’s take a look at some of the warnings. In 1902, future President Franklin D. Roosevelt met a Japanese student while studying at Harvard University who told him about Japan’s 100 year plan to take over Asia and the Pacific.
Famed aviation pioneer Billy Mitchell visited Japan in 1910 and concluded that “increasing friction between Japan and the U.S. will take place in the future there can be little doubt, and that this will lead to war sooner or later seems quite certain.”
Following WWI, Mitchell returned home and tried desperately to convince U.S. military commanders that one day, planes would be able to sink ships. He even demonstrated the power of aerial bombardment by sinking surplus Navy ships, and suggested Japan could use this capability against U.S. interests in the future.
In 1924, after touring U.S. Pacific military assets, Mitchell wrote “Winged Defense,” which which detailed how Japan could carry out an attack on Hawaii, predicting 100 bombers would strike Pearl Harbor. Not only did senior officers dismissed his warnings, but he was courtmartialed due to his continued demonstrations of aerial bombing.
CBS journalist Eric Sevareid, in early 1941, recounted how a young Korean American, who claimed to be part of an anti-Japanese Korean underground, came to his office and said “Pearl Harbor, before Christman.”
U.S. Ambassador to Japan Joseph Grew, considered by his contemporaries to have warm feelings toward the Japanese, learned of plans to bomb Pearl Harbor from the Peruvian Embassy in January 1941. Meanwhile, Secretary of State Cordell Howe dismissed the warning. In September, after receiving a report that the Japanese were told to “keep talking” until November, and that after November 29th, “things would start to happen.”
Then of course there were some key Japanese actions: Withdrawing from the treaty limiting its shipbuilding in 1936; singing of the tripartite pact with Germany and Italy in 1940; Japan’s military buildup.
Japan also vocalized key grievances regarding the United States. Japan cited a long history of aggression starting with Commodore Matthew Perry’s 1853 invasion of Japan; The exclusion act of 1924 which refused to let Japan join the league of nations and cut off all Japanese immigration to the United States. This act in particular angered the Japanese Emperor and other key leaders and they referred to the policy of an official policy of racial discrimination. Then there was the oil embargo in August 1941 which crippled Japan’s economy and ability to fight for its interests in the Pacific.
Finally, on the fateful day itself, a U.S. destroyer sank a Japanese submarine in the vicinity of Pearl Harbor. Radar indicated that Japanese planes were headed toward Hawaii but the watch commanders dismissed it as a glitch.
This is not an all encompassing list, but it is clear there were key pieces of information pointing to an attack. There were not only clear signs of intent, but clear signs of a significant military capability to actually deliver such an attack. However, all of these data points were dismissed.
So why were they dismissed? Human bias and a failure to connect the dots. Many political and military leaders did not believe the Japanese would attack because they dismissed Japan’s military power; many officials got sucked into the trap of mirroring — predicting adversarial intentions based on one’s own value system, not the enemy’s. They just could not fathom Japan doing something like this. As outlined in this article, there was not a single source of information, but a series of puzzle pieces that started coming in at the beginning of the 21st century. So there was an issue of connecting the dots.
Modern artificial intelligence and machine learning has the power to learn from surprise attacks like Pearl Harbor and aggregate all of these warning signs and present clear information of a pending attack. Aggregating, analyzing, and presenting this information in real time makes the apparent threat much easier to identify and allow decision makers to take the appropriate action. However, it's not foolproof. Humans still have to make the right decisions. In other words, we still have to deal with human bias. Whether it is an artificial intelligence system or a team of human advisors, if decision-makers let their own biases get in the way, they will still fail.
Allow me to drive home the point with one of my favorite lessons from history – the infamous Pickett’s charge from the battle of Gettysburg. General Robert E. Lee was convinced that a frontal assault would successfully crush the union forces. His most trusted subordinate commander, General Longstreet, warned him that it was a suicide mission. Longstreet approached Lee on three different occasions, offering an alternative course of action and expressing his belief that the attack would fail. General Lee still proceeded. Artificial intelligence, you see, is just another tool, a machine advisor. While it does a much better job of analyzing hidden data points, a human has to make a decision on whether or not to act on the information. It won’t matter how good the AI system is or what prediction it makes if humans are unwilling to consider the information and act in the appropriate manner.
Artificial intelligence has a lot of promise to prevent attacks like Pearl Harbor. But there are no guarantees unless humans overcome their own biases. That requires investing in a workforce with the acumen to leverage artificial intelligence.
