Reinforcement learning has mastered winning arcade games and the board game GO. More recently, DeepMind’s AlphaStar has mastered the more difficult video game StarCraft. Can we now safely extrapolate that AI will someday provide command strategy on the battlefield? Such a forecast is tempting. But command requires creativity and AI has yet to demonstrate creativity as defined by the Lovelace test. In military affairs, as in fields like medicine, AI can serve as an advisor, but in the most extreme cases will never be in charge.
To see the need for creativity in command, we need look no farther than the classic comic strip Calvin and Hobbes1. Calvin, drawing on his entire knowledge of the history of warfare, challenges Hobbes. He boasts “You see Hobbes, I have a water balloon and you don’t. I therefore have offensive superiority so you have to do what I say.”
Hobbes becomes creative outside of Calvin’s experience. He responds “I think I’ll take this stick and poke your balloon.” The final cartoon panel shows Calvin drenched with the water from his own balloon. He mutters “That’s the trouble with weapons technology. It becomes obsolete so quickly.”
There are many more serious examples of creativity from the history of warfare. In 218 BC, Roman military and naval might was supreme. Anyone examining the history of warfare would conclude that the Romans were unbeatable. But the Carthaginian commander Hannibal (247–183 BC) thought outside the box and famously led his forces, accompanied by elephants, across the Alps into northern Italy, thus taking the war directly to the Roman Republic. His creative move ranks among the most celebrated achievements of any military force in ancient warfare.
Nazi Germany invaded France in 1940 using the Blitzkreig, or lightning war tactic. Never before used, the strategy was unexpected and France fell.
Sun Tzu, author of the military classic The Art of War, notes that “all warfare is based on deception.” In military strategy, surprise actions deceive successfully when the enemy’s knowledge is based on past history, not on all present possibilities. For example, Pharnuches, one of Alexander the Great’s generals, had no experience with the mounted Scythian warriors who swarmed his army and he thus lost the battle.
When Cassius Clay (later Muhammad Ali 1942–2016 ) fought “Big Bear” Sonny Liston in 1963, boxing conventions featured opponents slugging it out nose-to-nose. Ali ignored the conventions. He boxed by eluding and taunting his opponents, thus exhausting them. The new creative strategy, which he called rope-a-dope, worked so well that at the beginning of the seventh round, Liston refused to leave his corner and continue the fight. He had no strategy to counter Ali’s originality.
Creative strategy wins wars. Fredrick the Great proclaimed “Everything which the enemy least expects will succeed the best.” It certainly plays a role in the technology that wins wars. In World War II, radar, the atomic bomb and the cracking of the Nazi code Enigma all helped shorten the war. But, as Calvin & Hobbes illustrate, less high-tech figures like Hannibal, the Scythian warriors, and Muhammed Ali could also be highly effective just using current technology if they were also using creative thinking that lies outside and beyond historical data. Such creativity is beyond the capability of AI.
Patrick D. Wall sums up the limitations of AI. “When you consider the great new ideas produced by men like Newton and Darwin and Galileo, you’ll find, initially, they had to throw away the old rules that they grew up with. Now machines do what they’ve been told to do. They obey the rules that have been fed into them by man. And we know of no machines at present that have means of overcoming this limitation.”
Dr. Wall was ahead of his time. His observation was made in 1961.2
1 Calvin & Hobbes, July 1, 1986
2 “The Thinking Machine” (1961) – MIT Centennial Film
Further reading: Thinking machines: Has the Lovelace test been passed? Surprising results do not equate to creativity. Is there such a thing as machine creativity?
Is data mining failing its first big test? Computers scanning thousands of paper don’t seem to be providing answers for COVID-19. (Robert J. Marks)